Terrestrial evaporation is associated with a turbulent proxy called ‘aerodynamic conductance’ (ga) through the land atmosphere exchange of sensible heat flux and it is coupled to the ‘canopy-stomatal conductance’ (gCS) through photosynthetic carbon exchange. While ga expresses the physical efficiency of heat and water vapor exchange, gCS controls the biological efficiency of vegetation canopy to gain carbon at the cost of transpiration. Thermal remote sensing diagnostics of evaporation from global Earth observation missions is based on solving the surface energy balance (SEB) equation. Scientists have long sought to capture the invisible evaporative flux from satellite land surface temperature (LST) by resolving the physics of turbulence in aerodynamic conductance and sensible heat flux, while estimating evaporation as a residual component of SEB. Somewhat surprisingly, the inclusion of canopy-stomatal conductance is overlooked due to the overriding emphasis on eliminating the uncertainties of this aerodynamic approach. However, the conductances are controlling factors of evaporation and only revising ga in an ad hoc manner can deteriorate model performances in heterogeneous landscapes. Capturing the two conductances using SEB theory is a crucial step for advancing evaporation monitoring and understanding the biophysical principles of ecosystem water use from thermal remote sensing observations. To test the theory, we explored the conductances of heat and water flux using the non-parametric Surface Temperature Initiated Closure (STIC) model and employed a combination of high spatial resolution LST from Landsat and ECOSTRESS along with AmeriFlux eddy covariance tower data across semiarid ecosystems in California. The heterogeneous vegetation and aridity gradient enabled us to investigate the consort of dual conductances at the flux tower sites. By pairing Landsat LST with Sentinel-2 multispectral observations and meteorological forecasts, we further inspected the ability of STIC to reproduce the effects of potential coalition of soil and atmospheric aridity on evaporation biophysics at the spatial scale. From flux tower measurements, we found novel results showing that the control of canopy-stomatal conductances on evaporative fraction operates in tandem with aerodynamic conductance, soil and atmospheric drought at all sites. While the synergy of meteorological forecasts and thermal remote sensing observations allowed visualizing and validating the dual conductances from space, repeating the spatial scale experiment at different times of the year confirmed how the impact of supply-demand limits of energy and water shaped the conductance feedback on evaporation. The analysis offers an alternative perspective combining physics with the explanation of biology to investigate the highly complex evaporation variability from space through future thermal satellite missions.

The spatial modeling has become an important tool to estimate evapotranspiration (ET) fluxes over regional and continental areas. One of the most widely used ET spatial models is the land surface temperature-based approach, as the land surface temperature (LST) is potentially a signature of both ET and the soil water availability via the surface energy balance. In recent decade, many efforts have been devoted to extract the LST from remote sensing data. Nevertheless, the spatial and temporal resolution of thermal-based remote sensing are coarse for the hydro-agriculture purposes. Therefore, there is a crucial need for LST data at higher spatial/ temporal resolution for monitoring the plant water status at the field scale. Consequently, downscaling techniques with different degree of complexity have been proposed to improve the spatial resolution of the LST data available at a high temporal frequency. The aim of this work is to develop a simple methodology to disaggregate LST and assess its impact on different LST-based energy balance models. TSEB (Two source energy balance model) and the PM (Penman Monteith model) models are used in this study. To do so, LST derived from MODIS (MODerate resolution Imaging Spectroradiometer) at 1 km resolution and six optical bands (Blue, Green, Red, NIR, SWIR) from Sentinel-2 at 20 m resolution have been selected. Multiple regression equation between the six optical bands and the LST has been applied to disaggregate the LST at 20 m resolution. The procedure is applied over 2 irrigated and 1 rainfed wheat fields in the Tensift basin, central Morocco. Overall, the results of disaggregated LST are very satisfying (R2=0.70, RMSE=4.78K). While comparing estimated ET by the two energy balance models with Eddy covariance measurements, the results over the three wheat fields are very promising, especially for PM model (R2= 0.63, RMSE=77 W/m2). The disaggregation approach should be evaluated over several crops, as it could help to assess the ET dynamics at high spatial resolution over the highly heterogeneous landscape in semi-arid agricultural areas, and will respond to the needs of managing agencies and help to better predict crop water requirements at the field scale as well as to manage the overall water resources in a sustainable manner.

In order to ensure farmers’ livelihoods in a changing climate, the optimization of irrigation and the diagnosis of soil water saving practices are required to tackle freshwater scarcity. Earth observation (EO) sensors provide synoptic and periodic coverage of spatially continuous radiometric measurements collected in a consistent, systematic, and objective manner. They are ideal for digitizing, and therefore better managing, plant water productivity and water use efficiency at different spatial scales: from farm to regional and national levels. In particular, a combination of thermal and optical observations can be used to estimate the actual soil evaporation (E) and canopy transpiration (T), usually referred to as evapotranspiration (ET). ET is a physical parameter, closely linked to plant development and health as plants both transpire and fix CO2 through the leaf stomata. Integrating remotely sensed ET and plant biophysical traits from different EO platforms would enable evaluating the water use efficiency in agriculture, not only in terms of carbon assimilation but, above all, on crop yield. These EO-based ET and crop productivity products at appropriate spatiotemporal scales would thus allow for optimizing irrigation allocation and provide diagnostic tools for soil conservation practices. In addition, the ET and crop yield estimates would become a vital information to monitor, mitigate and plan for the effects of prolonged droughts on food production. The EO MAJI project, as part of the EOAfrica Explorers initiative, makes use of inputs from the scientific ECOSTRESS and PRISMA sensors, together with sharpening and disaggregation techniques for fusing Sentinel imagery and physical models of plant biophysical processes, in order to produce daily ET and productivity estimates at farm scale. This will serve as an exploration of the capabilities of future operational satellite missions in deriving improved ET and yield/biomass products, as well as enhancing a better understanding of water use efficiency of cultivated landscapes. Both intermediate (ET) and final products (gross primary productivity and yield) will be evaluated against available in situ measurements in long-term experimental sites outside of Africa (Majadas de Tietar FLUXNET site in Spain and other ICOS sites), as well as in African sites in collaboration with African Early Adopters (Ministry of Agriculture of Burkina Faso and the Ministry of Lands and Water Affairs of Botswana). Together with CSIR and University of Pretoria of South Africa, we foresee to exploit existing crop yield and irrigation reports that are acquired by those institutions to validate project outputs.

Global climate projections expect an intensification of hydrological and agricultural droughts triggered by increasing temperatures in recent years. Due to changing climatic conditions, many areas are prone to deviations in water availability and water use that can be observed from evapotranspiration (ET) anomalies. Land surface temperature (LST) plays an essential role in regulating the exchange of water and energy between land and atmosphere, which directly relates it to ET. In this work, we exploit Sentinel-3 (S3) LST imagery for two-source energy balance (TSEB) modeling of ET at different spatial scales, including both original S3 dataset at 1 km spatial resolution and its downscaled products with 20-m and 100-m pixel size [Kustas & Norman, 1999]. In this regard, LST maps derived at multiple spatial scales together with multispectral Sentinel-2 (S2) instrument and ERA5 climate reanalysis dataset are used as main forcings for the TSEB model. First, we derive reflectance and biophysical parameters corresponding to different resolutions of the S2 datasets for thermal downscaling and then for ET estimation afterwards. In this context, we exploit relationships between 1-km LST grids and fine-resolution explanatory variables (both 20 m and 100 m) using decision trees algorithm [Gao et al., 2012]. Due to reduced capabilities of univariate models, multi-source predictors are considered, including time-coincident S2 observations to Sentinel-3 overpass along with slope and aspect derived SRTM DEM product. Consequently, meteorological forcings and solar radiation from ERA5 are generated for estimating instantaneous energy fluxes and daily ET at different spatial levels. This work will help to understand practical utility of the improved LST product for estimating TSEB outputs at higher spatial resolution and studying its performance against coarse-resolution data and in-situ measurements from eddy-covariance towers. Additionally, time-series of the improved LSTs are intended to temporally complement planned high-resolution thermal missions, including the ESA LSTM, TRISHNA, and SBG-Thermal as alternatives to the 1-km thermal observations from Sentinel-3 and MODIS instruments. Keywords: drought, two-source energy balance, Sentinel-3, land surface temperature, high-resolution thermal

Understanding terrestrial evapotranspiration over large ranges of time and space is critical for water resources management and climate change studies. The FAO Water Productivity Open-access Portal (WaPOR) uses the ETLook model to estimate near real-time and historical evapotranspiration data at field (20m), continental (100m) and global (300m) scale. ETLook calculates evapotranspiration using the surface energy balance and satellite-based inputs on NDVI, albedo and land surface temperature (LST). It is sensor-independent and can be applied at multiple spatial scales. However, the resolution and revisit times of current LST missions have some limitations for agricultural applications. Satellite missions such as LSTM, TRISHNA and SBG are expected to offer better spatial-temporal coverage from 2025 onwards. In the meantime, the spatial-temporal resolution of thermal infrared sensors can be enhanced with image analysis algorithms. A number of scientific publications describe the use of high-resolution optical data to sharpen thermal infrared imager (Mao et al., 2021). Gao et al. (2012) introduced the data miner sharpener (DMS) that builds regression trees between thermal infrared brightness temperatures and shortwave and optical reflectance values with additional information from digital elevation models. An implementation of this algorithm in Python is described by Guzinski et al. (2019). The DMS method is applied in conjunction with ETLook to produce evapotranspiration data at 100m for Africa and the Near East. The LST is obtained from VIIRS brightness temperature (band I5) at 375m, which is resampled to 100m using the DMS approach using Sentinel-2 and and elevation features. The resulting evapotranspiration data will become publicly available (May 2023) on the WaPOR portal and is the largest dataset so far for which the DMS approach is implemented. Although sharpening cannot replace actual thermal band imagery at high resolutions, it offers a good solution when no LST measurements are available for the resolution needed.

Evapotranspiration (ET) is a fundamental element of the hydrological cycle which plays a major role on surface water balance and surface energy balance. At local scale, ET can be estimated from detailed ground observations, for example using flux towers, but these measurements are only representative of very limited homogeneous area. When regional information is required, e.g. for monitoring ground water resources, ET can be mapped using thermal infrared and spectral reflectance data. Various ET models have been developed but there was no competitive evaluation of them over a large range of situations, so that it is not possible to evaluate the intrinsic performance of one model compared to another. In such situation, ensemble model averaging may be proposed for providing a coherent estimation of ET with an increased overall accuracy. In this work the ensemble modelling approach is extended to a multi-model – multi-data framework that provide ET estimations together with an uncertainty of estimation. We developed the EVASPA framework for estimating ET through ensemble averaging with the objective of providing estimates of ET together with an estimation uncertainty. In this presentation we examine three test cases. The first test presents a full analysis of the uncertainties of ET estimation in relation to uncertainties in input variables and models. Airborne remote sensing data were acquired over the Grosseto area in Italy in the frame of the ESA SurfSense experiment (high spatio-temporal Resolution Land Surface Temperature Experiment) in support of the LSTM mission project (Copernicus Land Surface Temperature Monitoring). This analysis shows that the main uncertainty sources are related to model formulation, solar radiation, wind speed and air temperature estimates. The second test case presents the monitoring of ET over Sahelian areas in the frame of the AMMA-Catch program using MODIS data. ET was estimated by considering weighted average of ET estimates by several models. The ponderation weights were proposed as a function of the a priori validity of each model depending on the season and the LAI level. The third test case presents an example of Bayesian model averaging over the Crau-Camargue test site in SouthEast France. ET estimates were obtained as weighted averages of several model estimations from MODIS data. The ponderation weights were obtained from each model evaluations against ground measurements of ET. For the two last test cases, ET was estimated with RMSE around 0.5 mm d-1. The EVASPA framework is presently used for the definition of the ET product in the frame of the TRISHNA thermal infrared space mission (CNES/ISRO).

Accurate estimates of terrestrial evapotranspiration are important for water resource management and improving water productivity in agriculture. The FAO Water Productivity Open-access Portal (WaPOR) uses the ETLook model to estimate near real-time and historical evapotranspiration data at field-scale (L3 20m), continental scale (L2 100m) and global scale (L1 300m). The ETLook model is sensor-independent and calculates daily evapotranspiration using the surface energy balance and satellite-based inputs on NDVI, albedo and land surface temperature (LST) at multiple spatial scales. This presentation will present the WaPOR version 3 (v3) evapotranspiration data which will be released in May 2023. WaPOR v3 has undergone significant improvements regarding the spatial and temporal quality of its inputs. One of the major limitations of the current v2 100m WaPOR dataset is the use of 1km MODIS LST to estimate relative soil moisture at 100m. The new v3 100m WaPOR dataset not only replaces MODIS LST by VIIRS brightness temperature at 375m but also applies a data miner sharpener (DMS) approach that builds regression trees between VIIRS LST and Sentinel-2 features to more accurately estimate LST at 100m. We will provide examples of the new evapotranspiration WaPOR v3 dataset at field, continental and global scale for different agro-ecological zones and also share some applications of evapotranspiration data. Furthermore, we will highlight the importance of future high spatial-temporal thermal infrared (TIR) sensors and address the requirements of these future TIR datasets to be used in an operational data processing pipeline.

Crop water stress (W) at a given growth stage starts to set in as moisture availability (M) to roots falls below 75% of maximum. It has been found that ratio of crop evapotranspiration (ET) and reference evapotranspiration (ET0) is an indicator of moisture adequacy and is strongly correlated with ‘M’ and ‘W’. The spatial variability of ET0 is generally less over an agricultural farm of 1-5 ha than ET which depends on both surface and atmospheric conditions while the former depends only on atmospheric conditions. Solutions from surface energy balance (SEB) and thermal infrared (TIR) remote sensing are now known to estimate latent heat flux of ET. In the present study, ET and moisture adequacy index (MAI) (=ET/ET0) have been estimated over two contrasting western India agricultural farms having rice-wheat system in semi-arid climate and arid grassland system, limited by moisture availability. High-resolution multi-band TIR sensing observations at 65m from ECOSTRESS (ECOsystemSpaceborne Thermal Radiometer Experiment on Space Station) instrument on-board International Space Station (ISS) were used in an analytical SEB model, STIC (Surface Temperature Initiated Closure) to estimate ET and MAI. The ancillary variables used in the ET modeling and MAI estimation were land surface albedo, NDVI from close-by LANDSAT data at 30m spatial resolution, ET0 product at 4km spatial resolution from INSAT 3D, meteorological forcing variables from short-range weather forecast on air temperature and relative humidity from NWP model. Farm-scale ET estimates at 65m spatial resolution were found to show low RMSE of 16.6% to 17.5% with R2 >0.8 from 18 datasets as compared to reported errors (25 – 30%) from coarser-scale ET at 1 to 8 km spatial resolution when compared to in situ measurements from eddy covariance systems. The MAI was found to show lower (0.5) magnitudes in the contrasting agricultural farms. The study showed the potential need of high-resolution high-repeat spaceborne multi-band TIR payloads alongwith optical payload in estimating farm-scale ET and MAI for estimating consumptive water use and water stress. A set of future high-resolution multi-band TIR sensors are planned on-board Indo-French TRISHNA, ESA’s LSTM, NASA’s SBG space-borne missions to address sustainable irrigation water management at farm-scale to improve crop water productivity. These will provide precise and fundamental variables of surface energy balance such as LST (Land Surface Temperature), surface emissivity, albedo and NDVI. A synchronization among these missions is needed in terms of observations, algorithms, product definitions, calibration-validation experiments and downstream applications to maximize the potential benefits.

Canopy temperature was proposed in the late 1970s as an indicator of water status in crops. Since then, various formulations and strategies have been developed that use thermal data to accurately quantify plant transpiration and monitor water stress for precision irrigation. Although most of the progress carried out in thermal remote sensing of crops in the last 30 years has targeted transpiration and water stress detection, recent research demonstrates the importance of thermal imaging also to detect early symptoms of plant stress induced by plant pathogens. This presentation focuses on the advances carried out in the past 20 years in thermal imaging combined with imaging spectroscopy for plant trait retrievals in the context of biotic and abiotic stress detection. In particular, results obtained with thermal and hyperspectral cameras on board aircraft and unpiloted platforms, along with physical models, will be discussed for the specific case of water stress detection and deficit irrigation, as well as for the early detection of diseases caused by harmful pathogens in agriculture and forest species. Results obtained with high-resolution thermal and hyperspectral images targeting tree crowns demonstrate the importance of thermal traits relative to those from hyperspectral-based indicators and the need for thermal imaging for the accurate detection of symptoms of stress at pre-visual stages. Recent work will be discussed, which shows the divergence of biotic-abiotic spectral fingerprints, and that the water stress-induced physiological changes can be separated from biotic-induced symptoms. The use of these physiology-based spectral traits linked to machine-learning algorithms will be discussed, as the existing gaps for the operational use of these methods for precision irrigation and biosecurity surveillance activities at larger scales.

Resolving the challenges of the future requires a thorough reconsideration of how water is managed in the agricultural sector, typically constrained by under-performance. Precision irrigation indeed represents an agricultural approach that supports the repositioning of the largest water consumer sector towards modernization and sustainability. Infact, exploiting the advent in remote sensing allows for an accurate and cost-effective assessment of the actual evapotranspiration (ETa) of the crops, thus of a precise estimation of their water requirements, and consequently supports the development of irrigation scheduling models and plans that improve water use efficiency. The project, funded by ESA within the “EO Africa Explorers” framework, is led by Planetek Italia and involves Planetek Hellas and CIHEAM-IAM Bari (International Center for Advanced Mediterranean Agronomic Studies - Mediterranean Agronomic Institute of Bari). The EO Africa project is supported by close interactions with local stakeholders playing the role of Early adopters. In particular, the Egyptian territory has been selected with the active involvement of the Egyptian National Authority for Remote Sensing & Space Sciences (NARSS) also engaged as EO Research Group of the project, and the “October sixth for agricultural projects” company. This project aims to develop and validate an open-source innovative model to assess actual ETa using EO-derived reference evapotranspiration (ET0), crop coefficient (Kc) and water stress coefficient (Ks). ​The solution will be integrated into a web platform as a Decision Support System (DSS) to improve irrigation water management.​ To develop the abovementioned products, data acquired by ECOSTRESS and PRISMA sensors will be used. ECOsystem Spaceborn Thermal Radiometer Experiment on Space Station (ECOSTRESS) mission, managed by NASA’s Jet Propulsion Laboratory (JPL), since 2018 provides Earth’s surface temperature from the International Space Station (ISS) at various times of the day with a spatial resolution of 70 m. Information in the TIR domain is used to assess actual evapotranspiration and soil moisture status. PRISMA is a hyperspectral mission of the Italian Space Agency covering the spectral range of 400-2500 nm in 244 bands. The PRISMA data are used to monitor the vegetation status and the generated products will be validated using several in-situ measurements in a large Egyptian agricultural area.

The retrieval of land-surface temperature (LST) from thermal infrared (TIR) remote sensing has been shown to be a valuable tool in surface energy balance models for estimating evapotranspiration (ET) However, it is difficult to monitor daily evapotranspiration with passive sensors such as Landsat, the revisit time (8 days) can be insufficient to detect changes in surface moisture or crop phenology, particularly in regions with persistent overcast conditions [1]. To address this challenge, the easy implementation of the common gap-filling methods, such as interpolating the ratio of actual to reference ET (ET/ETo) across image acquisition days, offers advantages in an operational context, but might not fully account for the non-linear dynamics of ET, especially over sparse-canopy irrigated crops. For this purpose, combining thermal and optical models can provide a more comprehensive and accurate estimate of crop evapotranspiration, by taking advantage of the operational use of dense time series of Earth Observation (EO) data by using different satellite platforms. In this study, we integrate the Dis-ALEXI model based on Landsat optical and thermal data [2] and the Shuttleworth-Wallace (SW) ET model based on Sentinel-2 optical data [3]. The aim is to evaluate the integration of the two frameworks and to determine the advantages of using high temporal observations in estimating ET instead of interpolated ones. The approach proposed in this study has been evaluated using flux tower observations in California vineyards and almond orchards, respectively in the GRAPEX (Grape Remote-sensing Atmospheric Profile and Evapotranspiration eXperiment) and the T-REX (Tree crop Remote sensing of Evapotranspiration eXperiment) projects. [1] Anderson, Martha C., et al. "Interoperability of ECOSTRESS and Landsat for mapping evapotranspiration time series at sub-field scales." Remote Sensing of Environment 252 (2021): 112189. [2] Knipper, Kyle, et al. "Evaluation of Partitioned Evaporation and Transpiration Estimates within the DisALEXI Modeling Framework over Irrigated Crops in California." Remote Sensing 15.1 (2022): 68 [3] D’Urso, Guido, et al. "Determining evapotranspiration by using combination equation models with sentinel-2 data and comparison with thermal-based energy balance in a California irrigated Vineyard." Remote Sensing 13.18 (2021): 3720.

Most models used to simulate Evapotranspiration (ET) forced with LST data, either in a forcing (SEB inversion) or a data assimilation (Land Surface Models LSM coupling SEB and water budget) are faced with the difficulty to relate the various key temperatures of the system. What is available from satellite remote sensing is the radiative surface temperature at the overpass time, while what is needed by most SEB models is the aerodynamic temperature representing the source temperature of the turbulent fluxes. Both temperatures are related to the distribution of the skin temperature of each elementary surface of the canopy. The radiative temperature is aggregating the longwave radiation from the elements in the field of view, thus depends on the view angle and azimuth; for a given solar illumination, the sensor will either sample more shaded or more sunlit elements (soil, leaves…). Whatever the surface radiative surface temperature might be, as resulting from these geometrical features, it should lead (through inversion or data assimilation) to the same ET value. On the other hand, the aerodynamic temperature will mostly result from the interaction between the eddies within and above the canopy, with more interaction with leaves for high LAI values and soil for the very low values of LAI. It is thus important to realistically simulate the main components of the skin temperature of the canopy. For preparation of the TRISHNA, which will have for any given pixel contrasted view angles from one overpass to the other, we investigated 1- whether we can improve the robustness of the ET retrieval from the single pixel SEB model SPARSE by accounting for the shaded/sunlit elements of the canopy, and 2- whether the distribution of skin temperatures simulated with a 3D SEB module in a LSM is realistic enough and the performance of simulating ET increases. For the latter, we compared the MAESPA pseudo-3D and the ISBA 1D models. Results show that 1- the stability of ET retrieval from one overpass to the other improves when extending the SPARSE model to 4 sources (shaded/sunlit soil and vegetation), with better partitioning between evaporation E and transpiration T but similar performances in ET retrieval for a wide range of canopies (wheat, oliveyard, vineyard), and 2- that although MAESPA simulates better ET, E/T and average canopy temperature than ISBA for a sparse rainfed oliveyard, the variability of LST is underestimated by MAESPA.

Land surface temperature (LST) indicates the thermal status of the surface as a consequence of the land-atmosphere exchange of energy and water fluxes. It is immensely sensitive to soil water content variations and evaporative cooling, and LST also carries the imprints of vegetation water use and stress. Since LST constrains the magnitude and variability of the surface energy balance (SEB) components, it serves as a pivotal lower boundary condition for retrieving evaporation (E) in thermal-based evaporation models. Through stomatal conductance, the transpiration component of E is intrinsically coupled with photosynthesis and captures the effects of water stress on photosynthesis. Therefore, LST closely relates to the terrestrial ecosystem processes. Taking advantage of the ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) observations, the European ECOSTRESS Hub (EEH) funded by European Space Agency (ESA) focuses on the water, energy, and carbon cycles in the terrestrial ecosystems. In EEH Phase 1 (2020-2022), we produced LST and instantaneous E data between 2018 and 2021 from models with different structures and parameterization schemes over Europe and Africa. LST was retrieved from the Split Window (SW) and Temperature Emissivity Separation (TES) algorithms. The retrieval of E was based on three models, namely the Surface Energy Balance System (SEBS) and Two Source Energy Balance (TSEB) parametric models, as well as the analytical Surface Temperature Initiated Closure (STIC) model. The evaluation showed that the SW LST had better performance over densely vegetated surfaces than the TES LST and the accuracies of these two LSTs were comparable in arid regions. E from the physically based STIC model had relatively better consistency with the measurements from the eddy covariance sites across varying aridity and diverse biomes. Taking the advantage of ECOSTRESS Collection 2 data, in EEH Phase 2 (2023-2025), we target at 1) refining the TES algorithm over vegetated surfaces by accounting for the cavity effect, 2) analysing the impacts of LST estimates from different algorithms (i.e., SW and TES) on E retrieval and the underlying aerodynamic and stomatal conductances over different biomes, 3) developing a hybrid look up table (LUT) approach for temporal integration of instantaneous E acquired at any time of day to estimate daily total E, and 4) coupling E with GPP by injecting the vegetation water stress information provided by E into photosynthesis. Map-ready time series (>5 years) of LST, daily E, and GPP are expected to be generated with public accessibility.

Thermal remote sensing can provide environmental measuring tools with capabilities for measuring both managed and natural ecosystem development and integrity. Recent advances in applying principles of nonequilibrium thermodynamics to ecology provide fundamental insights into energy partitioning in ecosystems. Ecosystems are nonequilibrium systems, open to material and energy flows, which grow and develop structures and processes to increase energy degradation. More developed terrestrial ecosystems will be more effective at dissipating the solar gradient degrading its exergy content. Terrestrial ecosystem's surface temperatures have been measured using airborne and satellite sensors for several decades. Using NASA’s Thermal Infrared Multispectral Scanner (TIMS) Luvall and his coworkers (Luvall and Holbo 1989; Luvall et al 1990; Luvall and Holbo 1991) have documented ecosystem energy budgets for including tropical forests, mid-latitude varied ecosystems, and semiarid ecosystems. These data show that within a given biome type, and under similar environmental conditions (air temperature, relative humidity, winds, and solar irradiance), the more developed the ecosystem, the cooler it's surface temperature and the more degraded the quality of its reradiated energy. These data suggest that ecosystems develop structure and function that degrades the quality of the incoming energy more effectively, that is they degrade more exergy, which agrees with the predictions of nonequilibrium thermodynamic theory (Schneider and Kay 1994a; Kay and Schneider 1994; Schneider; Sagan 2005 and Hamberg et al. 2020), This remote sensing work suggests that analysis of airborne collected radiated energy fluxes is a valuable tool for measuring the energy budget and energy transformations in terrestrial ecosystems. The ecosystem temperature, Rn/K*, Beta Index, and TRN are excellent candidates for indicators of ecological integrity. The same thermodynamic approach can be used for determining crop yield and optimum nitrogen fertilizer application. Alzaben (2020), using exergy destruction principle (EDP) tested under greenhouse and field conditions on corn plants at three different scales (i.e., leaf, canopy and over a plot area). Agricultural crops experiencing greater growth and providing greater yield will have lower surface temperature. Two hypotheses are developed as predicted by the EDP. It is hypothesized that agricultural crops experiencing greater growth and providing greater yield will have lower surface temperatures. The second hypothesis is that crops grown under optimum/higher rates of nitrogen will have lower surface temperatures compared to crops grown under nitrogen stress conditions.

Monitoring the temperature of forests is a key motivation behind upcoming high spatial and temporal resolution satellites such as TRISHNA, SBG and LSTM. The thermal regime over forests results from interactions at the vegetation, soil and atmosphere interfaces. Transpiration, evaporation and soil water- and heat transfer are the critical processes driving these interactions. The surface brightness temperature observed by satellites therefore depends on the forest radiative regime, and the emission, scattering and absorption of canopy and atmosphere irradiances. This includes their spectral and directional variations, resulting from shadowing effects, canopy-component temperatures and their distribution, and the sensor view angle. Directionality effects can therefore lead to significant biases over forests, whereby accurate and efficient schemes to understand and account for these effects are critical. Given the absence of multi-angular thermal remote sensing observations over forests, either experimental or modelling techniques have been used. From an observational approach, goniometric measurements of single forest components, thermal infrared (TIR) in-situ instruments such as radiometers or imaging cameras, and UAV/airborne flights have all been used to examine directionality in forests. However, these techniques may not be representative of the full canopy, may be subject to turbulence effects and do not measure angular effects at the satellite scale. Given that canopy structure may dominate directionality in different ways from the local to satellite scale, the causes and conclusions drawn from techniques at the local footprint may not scale up appropriately to the satellite footprint. Accordingly, this contribution aims to understand the causes of directionality at different scales over forests using an experimental approach. We use a goniometer and scale down the forest size to within the goniometer footprint whilst keeping the same structure observed at different scales. To test these assumptions, we a) increase the footprint size using different surfaces and b) select surfaces that are dominant in geometric or volumetric scattering. The results from this study will help understand the influences on directionality from tower to satellite scale, since local scale measurements are often used to infer directionality effects at the satellite scale.

Sea surface temperature (SST) is a fundamental physical variable for understanding, quantifying and predicting complex interactions between the ocean and the atmosphere. Such processes dictate how heat from the sun is redistributed across the global oceans, directly impacting large- and small-scale weather and climate patterns. The provision of daily maps of global SST for operational systems, climate modelling and the broader scientific community is now a mature and sustained service coordinated by the Group for High Resolution Sea Surface Temperature (GHRSST) and the CEOS SST Virtual Constellation (CEOS SST-VC). GHRSST SST products rely on a combination of low-Earth orbit infrared and microwave satellite imagery, geostationary orbit infrared satellite imagery, and in situ data from moored and drifting buoys, Argo floats, and Fiducial Reference Measurements (FRM) for product validation. The spatial resolution of accurate satellite SST observations has not changed dramatically in many years, but within a few years, several new satellites will be launched, that will provide accurate SST observations with spatial resolutions better than 100 meters. This new evolution will provide new opportunities for applications but will also require new developments within retrievals, validations cloud masking etc. It is therefore important that the new developments within high resolution SST products are coordinated with the ongoing international SST activities. Many global ocean SST users are requesting improved SST products near to and at the coastal zones; products with improved feature resolution including at frontal zones; and improved consistency of products at high-latitudes and at the marginal ice zone. At the recent GHRSST science team meeting, it was therefore decided to have the coordination of high resolution SST developments a priority for the coming years. In this presentation, we present the outcomes from the discussion among the GHRSST science team on the requirement for high resolution SST products and the developments that are needed. To make the maximum benefit out of the new SST products, it the importance of consistency between the traditional GHRSST compliant products and the high resolution products was stressed, as well as the importance of a dedicated SST retrieval algorithm and product in the product portfolios.

ECOSTRESS serves as a precursor for future high resolution thermal missions including TRISHNA, SBG, and LSTM. Lessons learned from ECOSTRESS may aid in the development of algorithms for the future high-resolution missions. Validation of ECOSTRESS skin temperature over the ocean was carried out using triple collocations among NOAA iQuam quality controlled in situ observations, geostationary SST and cloud retrievals and ECOSTRESS retrievals. Additional validation was carried out using Saildrone and shipborne thermistor and radiometer observations, which are in some cases collected at the ECOSTRESS pixel scale of 70 m. ECOSTRESS version 6 processing had a 1 K cold bias, which has largely been corrected in version 7.1 processing. Artifacts in the retrievals include striping and checkerboard patterns in regions where successive mirror scans overlap; these appear to originate from consistent non-uniformities in the sensitivity of pixels in the focal plane array that are not fully compensated by the L1 radiance calibration. Possible solutions include non-linear L1 radiance calibration (Arai & Tonooka 2005), detector-specific rather than band average lookup tables for radiance to brightness temperature mapping (Wang & Cao 2016), and adaptive destriping of L1 brightness temperatures prior to L2 processing (Bouali & Ignatov 2014). Despite the retrieval artifacts, ECOSTRESS is a game-changer for oceanography especially in the coastal zone, where it can resolve fronts, jets, filaments, mixing features and gradients previously undetectable by other satellite sensors.

The Indo-French TRISHNA (Thermal infraRed Imaging Satellite for High-resolution Natural resource Assessment) mission, foreseen in 2025, is designed to measure the visible, near infrared, and thermal infrared signal of the surface-atmosphere system globally and approximately twice a week at 60 m resolution for the continents and the coastal ocean. Thermal infrared imagery onboard land dedicated satellite missions - but also covering coastal areas - has been around since 1984 with Landsat series, offering today 100m resolution with a revisit time of the order of 10 days in mid-latitude. In coastal ocean, this wealth of data has been exploited only through limited and very localized demonstrations, mainly because of low revisiting time. Still they are invaluable for many applications ranging from forecasting system improvement to coastal process understanding and monitoring, and to the support to many economic activities such as aquaculture. Besides new missions (ECOSTRESS on ISS, TRISHNA in 2025, SBG in 2027, Copernicus LSTM in 2028) are now available or on their way, that will provide better radiometric performances, higher spatial resolution and temporal sampling and increase the value and operational capacity of this source of observation. Despite the wide range of coastal applications from such resolution satellite derived observations, there is no dedicated and validated coastal SST product from the existing missions distributed on regional or global scale. The primary objective of this study is to develop a state-of-the-art processing for ultra high resolution SST retrieval in coastal areas from the future TRISHNA mission. To achieve this objective and to fill the gap, the study aims at designing new 100m ultra-high resolution satellite-derived Sea Surface Temperature (SST) products dedicated to coastal waters from the existing Landsat-9 and ECOSTRESS imagery. By designing those new products (called “CALISTA”), the main challenges of the SST retrieval in coastal environment such as the land/water discrimination, possible modification of emissivity, interaction between high turbidity and cloud detection, algorithm optimization and validation are addressed. The preliminary results using ECOSTRESS data and the well-known split-window non-linear algorithm over the French coasts shows a bias of -0.02°C and -0.03°C and a standard deviation of 0.56°C and 0.60°C from matchups with VIIRS and Sentinel-3 SST data respectively. The results for both ECOSTRESS and Landsat-9 data will be presented.

Rising global temperatures are leading to urban heatwaves becoming longer, more frequent, and more severe, resulting in an increase in extreme heat exposure especially among vulnerable population groups. In the coming decades, global warming will only intensify these temperature extremes leading to an even higher likelihood of unsafe and deadly heat exposure conditions. Our best tool for studying fine-scale urban temperatures is through thermal infrared (TIR) remote sensing because we can quantify the magnitude of the surface urban heat island (SUHI) effect across all permutations of urban surface temperature gradients and complexity. Relying on air temperatures from ground stations alone is inadequate for representing fine scale temperature gradients due to their sparsity, and will usually lead to an underestimation of urban heating effects. The availability of TIR data at spatial resolutions of 100 m or less is generally required for distinguishing temperatures of different urban materials that can be made useful for urban planning. Multispectral thermal infrared data (TIR: 8-12 micron) such as from the ECOSTRESS mission launched in mid-2018, and upcoming TIR missions including LSTM, TRISHNA, and the NASA Surface Biology and Geology (SBG) in 2025-2029 will bring a golden age of high spatial resolution (< 100 m), multispectral TIR data, with a potential for a twice-daily global revisit. NASA’s SBG will include both a TIR and VSWIR instrument and is a core component of NASA's new Earth System Observatory (ESO) to improve our understanding of vegetation processes, aquatic ecosystems, urban heat islands and public health, snow/ice, and volcanic activity. In this study we explore the use of ECOSTRESS TIR data in urban heat science and applications. We will demonstrate the use of ECOSTRESS data to quantify the differences in the SUHI between heatwave and non-heatwave conditions during extreme heat events; for pinpointing hotspot locations in cities and quantifying the effects of urban heat mitigation interventions; and to estimate the heat index – a critical variable used to estimate the effects of humid heat on human health.

Land surface temperature (LST) is a key variable for urban microclimatology studies. The regular acquisitions of the upcoming thermal infrared (TIR) satellite missions - LSTM, TRISHNA and SBG - will allow for unprecedented investigation of the urban climate. However, retrieving accurate and comparable LST over cities remains a challenge due to the complexity of the observed surfaces. At the spatial resolution of these missions (37-60 m), the urban heterogeneity and 3D structure greatly impact satellite measurements. A detailed modelling of the 3D radiative processes of such complex scenes can help understanding which surface parameters should be considered in the LST retrieval algorithm and to quantify the uncertainties induced by methodological proxies or lack of accurate urban surface properties. The DART radiative transfer model is widely used to simulate radiative exchanges in the urban canopy along with corresponding remotely sensed images for any configuration (sensor, atmosphere, etc.). It is therefore a powerful tool to investigate the impact of urban surface on LST retrieved from satellite data. As there is no thermal computation in DART, the LST of each element of the scene needs to be provided as input. However, the large LST variability between the different urban elements makes a correct parametrization difficult. To overcome this issue, the idea is to chain DART with SOLENE-microclimat, an urban microclimate simulation tool coupling a thermo-radiative model for surface temperature calculation and CFD (Computational Fluid Dynamics) for airflow calculation. Among other variables, SOLENE-microclimat simulates LST at metric scale for each element of a 3D urban scene. DART can use this data to simulate satellite TIR images with a physically based LST distribution in the scene. This presentation gives an overview of the modelling chain approach and presents the first results obtained with synthetic and real data in link to the future IRT satellite missions.

Southern European functional urban areas (FUAs) are increasingly subject to heatwave (HW) events, calling for anticipated climate adaptation measures. In the urban context, such adaptation strategies require a thorough understanding of the built-up response to the incoming solar radiation, i.e., the urban energy balance cycle and its implications for the Urban Heat Island (UHI) effect. Despite readily available, diurnal Land Surface Temperature (LST) data does not provide a meaningful picture of the UHI, in these midlatitudes FUAs. On the contrary, the mid-morning satellite overpass is characterized by the absence of a significant surface UHI (SUHI) signal, corresponding to the period of the day when the urban-rural air temperature difference is typically negative. Conversely, nocturnal high-resolution LST data is rarely available. In this study, an energy balance-based machine learning approach is explored, considering the Local Climate Zones (LCZ), to describe the daily cycle of the heat flux components and predict the nocturnal SUHI, during an HW event. While the urban and rural spatial outlines are not visible in the diurnal thermal image, they become apparent in the latent and storage heat flux maps – built-up infrastructures uptake heat during the day which is released back into the atmosphere, during the night, whereas vegetation land surfaces loose diurnal heat through evapotranspiration. For the LST prediction model, a random forest (RF) approach is implemented. RF results show that the model accurately predicts the LST, ensuring mean square errors inferior to 0.1 K. Both the latent and storage heat flux components, together with LCZ classification, are the most important explanatory variables for the nocturnal LST prediction, supporting the adoption of the energy balance approach. In future research, other locations and time-series data shall be trained and tested, providing an efficient local urban climate monitoring tool, where in-situ air temperature observations are not available.

Over the last decades, the frequently reported heat waves due to climate change have taken a heavy toll on the human health and food production of a densely populated country like India. Traditionally 2 m-air temperature is used for monitoring the onset, severity and withdrawal of heatwaves. However, the air temperature stations are sparsely located and using high spatiotemporal Land surface temperature (LST) can be more effective in monitoring heatwave events. In this study, we analyzed the spatial and temporal patterns of Land surface temperature over Jalgaon, a district in Maharashtra where frequent heatwaves were reported in the month of April for multiple years. First, the standardized LST anomaly maps were prepared using the observations from the MODIS 8-day average LST product for the month of April. Based on the LST anomalies, we identified the years 2017 and 2019 as cooler and warmer years respectively. Then, to further study the LST patterns, we disaggregated the multi-time MODIS Terra and VIIRS (day and night) LST images from 980 m to 70 m using a multi-variable disaggregation technique, which was already tested and validated over several sites. Later, the diurnal temperature cycle (DTC) at 70 m was characterized over urban, cropland, and bare soil pixels by fitting the disaggregated LST observations using a DTC model, GOT01-ts. The results indicated a significant increase in the minimum temperature (around dawn), maximum temperature (noon), and temperature amplitude for the year 2019. The night time LST was also higher during 2019 for all three land cover types. It was observed that time at which the maximum LST in a day occurred about 30 minutes earlier (~13.30 local time) during heatwave year than cooler year (~14.00 local time). Further, during 2019, the amplitude of the LST cycle over cropland exhibited a significant increase compared to 2017 indicating heat stress on vegetation. Overall, the diurnal cycle of LST was more pronounced during heatwave days than on normal days. The disaggregated diurnal LST can clearly indicate the effect of heat stress at much finer spatial scales and further research is needed towards an operational heatwave monitoring system.

Snow, as a fundamental reservoir of freshwater, is a crucial natural resource. Specifically, knowledge of spatial and temporal variability of snow density could improve modelling of snow water equivalent, which is relevant for managing freshwater resources especially in context of ongoing climate change. Some attempts to retrieve snow density were conducted using active and passive microwaves, but the possibility to get accurate estimates of snowpack density from remote sensing still represents a great challenge. In this contribution, we present an innovative method that combines satellite optical and thermal Landsat-8 images, meteorological parameters, snowpack modelling and field snow measurements to estimate thermal inertia of snow in Alpine basins sited in Italian Alps. Thermal inertia is then used as indicator of temporal evolution of snowmelt processes and to estimate snowpack density at catchment scale. Thermal inertia represents the responsiveness of a material to variations in temperature and depends on the thermal conductivity, density and specific heat of the medium. Thermal inertia has been successfully exploited in various applications regarding surface planetary geology, urban heat island and soil moisture detection, but it is never exploited to evaluate snow properties, such as snow density. Results provide evidences that snowmelt phases can be recognized in time and that bulk snowpack density can be estimated from thermal inertia observations. The developed semi empirical regression model (between thermal inertia and snow density field measurements) was applied at catchment scale to demonstrate the possibility of using optical and thermal data to estimate snowpack density. The model allows for estimation of snow density with a cross-validated coefficient of determination and cross-validated root mean square error of 0.59 and 82 kg/m3, respectively. Overall, this study shows the possibility of exploiting snow thermal inertia for snow density monitoring and it may open new frontiers in the remote sensing of the cryosphere.

The snow surface temperature in mountainous areas is largely variable in time and space because of multiple complex topographic features. The CNES/ISRO satellite Trishna will provide surface temperature measurements with an unprecedented combination of high resolution (60 m) and short revisit time (3 days) starting late 2024. Based on numerical experiments, we anticipate that these data will significantly improve the estimation of the snow water equivalent, the key variable in snow hydrology. However, our knowledge of the processes controlling the variability of the snow surface temperature still has gaps that could impede the assimilation of Trishna observations. Also, the future assimilation of actual Thrishna observations will require a good knowledge of their accuracy and of the physical processes controlling their subpixel variability. In the framework of the mission preparation, we built a large dataset of surface temperature measurements from fixed and UAV-mounted TIR cameras and TIR radiometers at the Col du Lautaret site in the French Alps. Surface temperature maps obtained using the energy-balance based RoughSeb model were then compared to in-situ measurements, as well as to Landsat 8 and 9 Land Surface Temperature products to assess the performance of the model. Here we present the results of this evaluation, including assets and limits of the RoughSeb model and showing its potential for the evaluation of satellite surface temperature products on snow-covered areas over complex terrain.

Over the years, FAOs WaPOR dataset containing information on, among others, evapotranspiration and biomass-production has proven to be useful in assessing water productivity in agriculture and water accounting studies. With pyWaPOR, it is now possible to generate data similar to that found in the WaPOR database outside of the region for which WaPOR data is available (i.e. currently Africa and the NENA region). By giving users access to the models (WaPOR-ETLook, C-Fix and SERoot) developed by the FRAME Consortium for the WaPOR methodology, they can access intermediate parameters, ingest data from different sensors into the models and customise the models to meet their specific demands. Besides the models themselves, pyWaPOR also includes a range of tools to prepare data from different sources for ingestion into the models. These tools allow for automatic downloading, reprojecting, merging, gap-filling, temporal-interpolation, compositing and more of datasets. PyWaPOR is written in Python, open-source, documented and can be installed through pip and run in a Colab notebook. PyWaPOR is being used, among others, in the Indus basin in Pakistan to support improved irrigation management. The presentation will highlight the role of sharpened high resolution, multi-sensor thermal data in pyWaPOR and its application in FAO programs

One of the main applications of satellite derived land surface temperature (LST) data is the modelling of actual evapotranspiration (ET) of crops with the purpose of monitoring and improving irrigation practices and crop water use productivity. Evapotranspiration is a highly dynamic process, both in time and in space, and therefore it requires LST observations with high spatio-temporal resolution. None of the currently operational spaceborne thermal sensors can fulfill this requirement and therefore data fusion between various optical and thermal sensors is often employed to try and bridge this data gap. Previous studies demonstrated the utility of fusion of shortwave-optical Sentinel-2 observations with thermal Sentinel-3 observations to derive daily, field-scale ET estimates. However, those studies also demonstrated the limitations of the approach in capturing the sharp thermal contrast between the cooler LST of recently irrigated agricultural parcels and surrounding hotter dry areas. In the current study we attempt to address this limitation by including information on the thermal spatial variability observed by Landsat satellites into the data fusion process. The methodology is designed in such a way as to conserve the thermal energy of the Sentinel-3 observations at their native resolution and not to be limited by infrequent and/or cloudy Landsat thermal observations. In addition, in previous studies the evaluation of ET derived with data-fusion sharpened LST was mostly performed in semi-arid Mediterranean climate. In the current study, we extend the evaluation to other, cloudier, climates. The results will guide the Food and Agricultural Organization in future developments of their WaPOR portal and will provide a basis for further development of thermal data fusion techniques incorporating new generation of thermal sensors, such as LSTM.

Agriculture will progressively require more and more attention as changing climatic conditions and reduced water availability threaten food security worldwide. The optimization of the agricultural production is obtained with constant monitoring of the plant health (in terms of e.g., soil moisture, leaf temperature or evapotranspiration), which can be challenging if crop fields are too extensive. Thermal observations from remote sensing are extensively used in agricultural monitoring to power (mostly-residual) energy balance model that provide evapotranspiration estimates. Two main issues hinder the quality of the results from these models: (a) sub-pixel heterogeneity, in particular related to mixed crops (e.g. row and tree crops), which can be captured only partially by the available LST information and (b) temporal frequency of the information, which for most freely-available products is usually at odds with spatial resolution (e.g., 1 km data from MODIS is available daily, whereas 90 m data from Landsat only once every 7-8 days). Furthermore, tree crops draw water from deep layers of soil, further disconnecting the satellite information from the biophysical processes involved in plant growth. In this work, the use of a continuous, two-source, double-soil-layer coupled energy-water balance model is displayed as a solution of these issues. The link between the two balances allows to compute surface temperature internally, meaning that satellite LST observations are used, only when available, for the calibration process. Furthermore, the use of a double source in the energy exchanges allows to properly address the intra-pixel heterogeneity. Finally, the double soil layer allows to address the soil water and energy vertical gradient in complex systems, properly framing the surface observation from remote sensing within the overall environment.

Vegetation in all of its forms (i.e., grasslands, agriculture, forests, steppes, etc.) is considered an essential pillar, as it provides almost all necessities to ensure life continuity on earth. However, due to climate change, the degradation of this substrate is becoming more and more evident, which may cause serious imbalance and irreversible damage to our ecosystem. Thus, preserving this natural grace is very significant, and one of the valid methods to achieve this, is observing its variations over time by employing remote sensing imagery (i.e., thermal, visible, SAR, etc.). In this study, a method to enhance monitoring vegetation will be used, which consists of finding a compromise between four related remote sensing biophysical indices (i.e., Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Atmospherically Resistant Vegetation Index (ARVI), and Optimized Soil Adjusted Vegetation Index (OSAVI)) by computing the average time series of their time series. This method was applied to San Severo-Italy’s quarterly very high-resolution Sentinel-2 thermal and visible images of 8 years (2015-2022), where it first resulted in NDVI, EVI, OSAVI, and ARVI’s time series featuring the same annual cycle and trend, with slightly different levels. And by using the average method, a multitemporal graph was obtained ranging between 0.15 and 0.5, and combining the four biophysical indices and their characteristics, while preserving the region’s seasonality and trend, with a standard deviation ranging between 0.02 in summer and autumn seasons and 0.09 in winter and spring seasons. Obtained graph shows that the seasonality of vegetation in San Severo from 2015 to 2022, peaks in spring season and declines to its lowest values in autumn. In addition to its trend decrease by 18%.

Within “ARIES”, experimental EO analysis techniques will be developed and validated, addressing water management and food security in Africa. These techniques, algorithms and prototype solutions will be based on a new generation of operational EO data from thermal (ECOSTRESS) and hyperspectral (PRISMA/EnMAP) satellite sensors. More specifically, we will investigate the synergy between these new data sources and operational Copernicus data services (mainly Sentinel-2 and Sentinel-3) to generate high-resolution indicators on crop growth and water stress. As such, the experiences gained within this project will deliver important information for the design of future Copernicus missions (CHIME, LSTM). The project is specifically centered on Africa with the intention to exploit Earth observation data for societal needs in Africa. “ARIES” aims to create more detailed and timely information about drought conditions and crop water stress for African land use stakeholders. Thus, helping them navigate changing climatic conditions with unreliable rainfall patterns, that threaten food security. On an individual field or farm level this could e.g., take the form of more timely irrigation advice. On a larger scale the information that will be generated aims to inform drought policy frameworks in the respective regions. To ensure the products developed within the project serve the needs of future users, we developed an integration strategy with five African Early Adopters. These partner organizations and their designated test sites are covering different regions in Africa as well as different agricultural management systems (irrigated and non-irrigated croplands and pastoral systems). Thereby, the developed algorithms and approaches can be validated, tested and evaluated in different geographical regions with different climatic conditions and agricultural practices. The algorithms will be implemented on the already existing online platform “Food Security TEP”. This is also where all prototype data and algorithms will be published at the end of the project. At the workshop, we will show the set-up of the project as well as the current status of the user requirements definition and algorithm development. The project runs from October 2022 until April 2024 and is funded by ESA (ESA Contract No: 4000139191/22/I-DT).

African farmers are facing the challenges of a changing climate, increased temperatures, changes in rainfall patterns, more frequent extreme weather events and reductions in water availability. The digital transformation of the agricultural sector is one of the opportunities that can promote good practices of the African agricultural through the sharing of information and tools for decision-making, thereby, boost economic growth of our African country. The shift to digital technologies is expected to move the sector from resource-intensive agriculture toward precision farming, helping it respond as much to the demands of market competition as to the challenges of adapting to climate change. In this context, Morocco has started in 2020 an ambitious agricultural plan called "Green Generation" in 2020 to rely more heavily on digital transformation which can help to reduce the pressure on its fragile resources of water and land degradation. To ensure a profitable agricultural sector in Morocco and to make informed decisions, it is important to understand the state and trends of agricultural production and deliver cost-effective, timely and accurate methods to better support the need of annual crop information. Until recently, digital agricultural crop mapping at the Moroccan national scale has been a challenging, especially with regard to data collecting, storing and processing and the requirement of the datasets to cover large geographic areas. To this end, this study responds to the urgent need for annual crop inventory to be made available following the growing season. This study will take advantage of the ever-increasing availability of high-resolution open-access Earth Observation (EO) data at both optimal spatial and temporal scales and powerful computing resources to achieve agricultural mapping applications. Such information can be used to better support Moroccan policies, programs, performance measurement and to address key environmental challenges along the country sustainable development goals. For the purpose of this study, we will use EO imagery (e.g., Sentinel and Landsat imagery), advanced modelling algorithms and other monitoring systems to produce operational agricultural annual crop inventories as well as crop type digital maps for Morocco. To do so, we aim to collect extensive ground observation data over two basins (i.e., Oum Errabi and Tensift) and other monitoring systems to provide information relating to agricultural production, develop new machine/deep learning classifier algorithms, develop fully automated crop classifier that should significantly reduce digital production time for subsequent years. In a first stage, the produced crop digital maps will be validated and calibrated for the two selected basins with a minimum target accuracy (i.e., >80%); then in a second stage these mapping activities will be extended/upscaled to most of the Moroccan national scale. By producing a Moroccan annual digital crop map inventory, ultimately we aim to (i) provide high quality information on the location, extent and changes of Moroccan crops, (ii) have an impact on the Moroccan agriculture sector and beyond, (iii) constitute an important foundational data source for a number of activities. Such information can be used to better support Moroccan policies, programs, performance measurement, (iv) support several key environmental indicators, and (v) understand changes in the environment over space and time, since the EO Data given by the inventory are spatially and temporally referenced. Preliminary results will be presented during the INTERNATIONAL WORKSHOP ON HIGH-RESOLUTION THERMAL EO

Evapotranspiration (ET) is a key variable in the understanding of the hydrological cycle. However, in many regions, like the Sahelian Regions, there is a spatial scarcity of in-situ ET measurements, in spite of its vulnerability to water availability and food security problems. However, in the past decade, many spatialized ET products have been released. They use various calculation methods like physical or empirical modelling, upscaling of in-situ measurements or data fusion approaches. The aim of this study is to propose a quite exhaustive review and evaluation of global or continental ET products available over typical Sahelian ecosystems in both Senegal and Niger, in the frame of the EVAP’EAU project (ICIREWARD Unesco Center). 20 ET products have thus been evaluated at local scale, using flux tower measurements over a typical agropastoral ecosystems. A meso-scale (~150km) evaluation has also been performed, by doing a cross comparison of the products at different spatial aggregation levels. Results show that the products with the best temporal representation of ET have the lowest spatial resolution (>10km), and thus lack of spatial representativeness. On the other hand, higher resolution products (<1km) show a realistic spatial distribution but several issues on the representation of the ET cycle seasonality. Therefore, in order to tackle water and agricultural management issues, there is need for better spatialized ET estimates at both high spatial and temporal resolution in Sahelian region. This could be achieved by proposing new data fusion methods, dedicated to these issues. However, the upcoming TRISHNA (CNES-ISRO), LSTM (ESA) and SBG (NASA) satellite missions will help to fill this gap by providing TIR data and products with high spatial (~50m) and temporal (~2 days) resolution.

The diurnal cycle of evapotranspiration (ET) provides information on the physiology of vegetation and other related physical processes. The observation or modelling of diurnal ET is restricted to in situ measurements due to the unavailability of diurnal Land Surface Temperature (LST) which is an important input to ET models. Multiple studies are done to get high spatial resolution LST and ET. However, studies on getting high spatiotemporal LST and modelling diurnal ET are rather limited. Diurnal cycle of LST can be observed from thermal sensors in the geostationary orbit at a coarser spatial resolution. The multi-time observations from polar orbiting sensors such as MODIS combined with a diurnal temperature cycle (DTC) model can also provide the diurnal cycle of LST at spatial resolution in the order of 1 km. We have developed a disaggregation approach to get field scale (at 70 m) diurnal cycle of LST by combining the multiple MODIS/VIIRS observations with a four-parameter DTC model. The objective of the study is to compare these diurnal LST at different spatial resolutions to model the diurnal cycle of ET. The study was carried over a vineyard and a crop land in India using two ET models- STIC and PT-JPL. Apart from LST, all necessary inputs to the ET models were obtained from in situ observations. On comparing against ground observations, the disaggregated diurnal cycle of LST at 70 m was found to be better than 1 km observations from MODIS or the 4 km observations from the geostationary satellite INSAT indicating the improvements brought out by the disaggregation model. However, the 1 km MODIS LST observations and the 70 m disaggregated LST resulted in similar values and patterns of diurnal ET for both the models and sites. The RMSE of the diurnal ET was similar at 1 km and 70 m, however better than 4 km INSAT observations. The results indicate that the improvements in LST due to disaggregation is not getting propagated into ET modelling. These results suggest that further research should be carried out to improve ET models for better monitoring of diurnal ET at field scales.

Evapotranspiration (ET) is a key input for irrigation scheduling and crop yield forecasting. Especially for high-value crops such as grapes, plot scale ET values are required by farmers for precision irrigation and maintaining prescribed levels of water stress in vineyards. Though remote Sensing-based ET modelling has been advancing with the availability of higher spatial and temporal resolution optical data, estimating ET at plot scale over vineyards is difficult considering the spatial heterogeneity of different surface variables and land surface fluxes. In developing countries such as India, critical data for accurate ET modelling over vineyards are often lacking and it is necessary to use simpler ET models. Further, the spatial resolution of current generation thermal infrared sensors is not enough for plot-level ET monitoring; hence, spatial disaggregation of ET from thermal sensors becomes important. The aim of this study is to assess if we can monitor ET over vineyards at plot scale towards water management during the cropping season. As the first step, a two-source ET model Soil Plant Atmosphere and Remote Sensing Evapotranspiration (SPARSE) was used for modelling ET and secondly, a contextual disaggregation model was used to obtain ET at 3 m resolution. The ET modelling was carried out using Landsat-8 and Plantscope datasets over vineyards in Ripperdan Ranch, Madera, California and Malegaon district in India. The RMSE in the SPARSE model ET was 1.19 and 1.41 mm/day over the sites in USA and India respectively. The spatial diaggregation resulted in marginal improvements in the accuracy of the ET with finer spatioal variations being captured. However, the accuracy in estimating the total crop seasonal ET depended on the number of clear sky images over the site. Over the site in USA, where multiple clear sky images were available, the sesonal ET was obtained within 15% of the in situ observations. However, over the Indian site where only limited images were available, seasonal ET was significantly overestimated. This calls for high repeat cycle thermal sensors in space for improved water management.

In the framework of the ESA's efforts to support the EO4AFRICA community by implementing initiatives that encourage the adoption of Earth Observation products following an African user driven approach, the present consortium is working within the “EO AFRICA EXPLORERS – PRISMA 4 AFRICA” project for the combination of hyperspectral data (i.e. PRISMA) and thermal data (i.e. ECOSTRESS) as precursors of the CHIME ESA mission and the ESA LSTM and the NASA-ASI SBG-TIR missions. Based on our first interactions with African Early Adopters, sugarcane has been identified as crop of their interest for a combined use of thermal and hyperspectral EO data. As preliminary use case, we started to set up a monitoring model on a sugarcane test site located in Iran for which in situ ancillary (e.g. irrigation timing, weather stations and in situ LAI and pigments measurements) and VAL data were available. The methodology will then be exported to the African countries involved in the project. ECOSTRESS is exploited to study the plant's water stress by utilizing L3 (ET-PT, ET-ALEXI) or L4 (e.g. WUE), while PRISMA is applied to retrieve sugarcane biophysical variables related both to structure (e.g. LAI, fPAR and FCOVER) and pigments (e.g. LCC and carotenoid contents) as well as crop stress indicators as mimic by the PRI and others ad hoc narrow bands spectral indexes. Moreover, we are investigating the possibility to apply PRISMA images to derive the ancillary information required by ECOSTRESS processing chain for ET calculation. Even though LST products are available and of good quality, the lack of such ancillary data prevents ET products to be generated. The PRISMA ancillary information (i.e., albedo and LAI) were used to configure the SEBAL Rcode input (https://rdrr.io/github/gowusu/sebkc/man/sebal.html) to derive potential and actual ET products. Tests have been also performed on contemporary Landsat/PRISMA acquisition showing an R2=0.94 for LANSDAT ET standard product vs Landsat/PRISMA combined product (SEBAL algorithm). Preliminary results show that the ET 70m products derived combining PRISMA and ECOSTRESS are of good quality in terms of dynamic range and spatial pattern so that they could be applied to better describe the phenological growing cycle of the sugarcane crop filling the gaps of the ECOSTRESS L3 products availability.

Water scarcity and the inter-annual variability of water resources in semi-arid areas are limiting factors for agricultural production. Characterization of plant water use, together with water stress, can help us to monitor the impact of drought on the agro- and ecosystems. It is especially true in Sahel region as it is identified as a « hot spot » for climate change. In such regions, in-situ measurements are often insufficient to accurately assess the variability present in the study area due to the sparsity of gauges networks. To tackle this issue, remotely sensed evaporation estimates, derived from thermal infrared data can be used. In this study, spatially-distributed estimates of daily actual evapotranspiration (ETd) are simulated using the EVASPA S-SEBI Sahel (E3S) model, which is based on the Simplified Surface Energy Balance (S-SEBI) contextual method and the EVapotranspiration Assessment from SPAce (EVASPA) tool. Such contextual approaches assume the simultaneous presence of sufficient fully wet and fully dry pixels within the same satellite image. E3S uses a set of different alternative methods in order to identify these limit conditions, called dry and wet edges, on the surface temperature/albedo scatterplot and consequently the Evaporative Fraction (EF) of each pixel in the image. However, this assumption is not always true, especially in the Sahel which is characterized by a strong seasonal climate contrast, due to the West African monsoon. To address this issue, we provide a sensitivity analysis to assess the effect of using different EF estimation methods over different spatial coverages. The work presented in this study allows to identify adapted methods for a correct determination of wet and dry edges in both highly dry and highly wet images. E3S was applied with MODIS/TERRA and AQUA thermal infra-red and visible datasets in the Sahel region. From this analysis, a procedure of methods selection according to the heterogeneity conditions is proposed, for an operational application in the future Indian-French high-resolution thermal mission Trishna (60m, 2days).

Thermal emission from the crop canopy is a sensitive parameter with respect to its vigor / stress, which influences the partitioning of energy and mass fluxes at the earth surface. Canopy temperature derived from high resolution satellite based thermal data can have multiple local scale applications like detection of crop stress, estimating evapotranspiration (ET), energy balance studies, gross primary productivity estimation, etc. The present study has been conducted in the parts of Ujjain district, Madhya Pradesh, India to evaluate the LWIR band of an experimental high resolution thermal satellite data (HRT) for assessing the crop stress in combination with the optical Sentinel-2 and LISS-4 data. The object-based delineation of field boundaries was carried out using multi-resolution segmentation applied on the canny edge layer derived from PAN channel of HRT and three bands of LISS-4 data with optimal segmentation weights and scale. Wheat crop were classified using multi-band Sentinel-2 data acquired closest to the HRT acquisition. LST were generated from brightness temperature of LWIR band of HRT data using single channel technique with emissivity derived from Sentinel-2 NDVI. HRT-LST and Landsat-8 LST of nearest date was found to be highly correlated, proving the data quality of the HRT. Scatter plot derived from LST (LWIR) and Sentinel-2 NDVI was used to generate the dry and wet-edges equation to derive Temperature-Vegetation Dryness Index (TVDI). TVDI computed for wheat polygons showed marked variations across the study area. A significant difference in the TVDI values of healthy wheat plot and stressed wheat plots were observed when correlated with the ground observations. The study showed the potential of high resolution thermal data for local scale crop stress detection. Such product can successfully be utilized to crop yield modeling, ET & GPP estimation, irrigation scheduling etc.

The use of land surface temperature (LST) for monitoring the consumption and water status of crops requires data at fine spatial and temporal resolution. Unfortunately, the current spaceborne thermal sensors provide data at either high temporal (e.g., MODIS: Moderate Resolution Imaging Spectro-radiometer) or high spatial (e.g., Landsat-8) resolution separately. Disaggregating low spatial resolution (LR) LST data using ancillary data available at high spatio-temporal resolution could compensate for the lack of high spatial resolution (HR) LST observations. Nevertheless, existing downscaling approaches do not consider soil water availability to explain the variability in LST at HR. In this context, a new method is developed to disaggregate kilometric MODIS LST at 100 m resolution by relying on Sentinel-1 (S-1) backscatter, which is indirectly linked to surface soil moisture and Sentinel-2 (S-2) reflectances, which characterize the green and total vegetation covers. The approach is tested over an 8 km by 8 km irrigated crop area in central Morocco (Marrakech) on the dates when S-1, S-2, and Landsat-7 or Landsat-8 acquisitions coincide with a one-day precision during the 2015-2016 growing season. The approaches are first applied to the 1 km aggregated Landsat LST as an initial step. Then, the 100 m disaggregated LST is compared to Landsat LST in three cases: no disaggregation, disaggregation using a green vegetation index (NDVI) derived from S-2 data, and disaggregation using both S-2 NDVI and S-1 backscatter. When including S-2 NDVI only in the disaggregation process, the root mean square error in LST decreases from 1.87 to 1.37 °C and the correlation coefficient (R) increases from 0.72 to 0.94 compared to the non-disaggregated case. The new methodology including the S-1 backscatter as input to the disaggregation is found to be more slightly more robust on the available dates with a disaggregation error decreasing to 1.30 °C and an R increasing to 0.95. As a second step, these approaches will be also tested using the 1 km resolution MODIS data as input.

Land surface temperature (LST) is an essential input variable for various environmental and hydro-meteorological applications including crop growth monitoring, irrigation need, and yield estimation. Crop monitoring requires high repetition frequency with high resolution LST data to detect the change in hydric condition, as water stress may occur throughout the growing season, especially in arid and semi-arid areas. Remote sensing observation offers the possibility to estimate the LST in the spectral range of thermal infrared (from 8 to 14 µm) with various temporal and spatial resolutions. Nowadays, the present satellite thermal sensors offer a trade-off between temporal and spatial resolution. Some sensors, such as Landsat and the Advanced Spaceborne Thermal Emission Reflection Radiometer (ASTER), have a high spatial resolution (100m) but a low temporal resolution (16 days), while others, such as the Moderate Resolution Imaging Spectroradiometer (MODIS), has a high temporal resolution (daily) but a lower spatial resolution (1km). In this context, disaggregation of low spatial resolution LST seems a great alternative to improve the spatial resolution of LST products. This work aims to disaggregate MODIS-LST 1 km to 100 m by combining Sentinel-1 and 2 data with machine learning algorithms over a semi-arid area characterized by its heterogeneity in terms of soil conditions and crop specie. Four machine learning algorithms were tested in this study including, Multiple Linear Regression (MLR), Support Vector Machine (SVM), Radom Forest (RF), Long Short-Term Memory (LSTM). The results show that the SVM method provides more robust and accurate results for LST disaggregation with a correlation coefficient (R) of 0.82 and a Root Mean Square Error (RMSE) of about 1.54 °C between disaggregated LST and Landsat data LST. The disaggregated LST will be incorporated into a combination of the energy balance and light use efficiency models for crop water needs and yield estimation in the study areas.

Sensors from Low Earth Orbit (LEO) can acquire multi- and hyper-spectral radiance data from which L2 data can be derived. Converting acquired raw datasets into science data is complex and often requires extensive computational capabilities, which are currently not available on-board most satellites, especially cube-sats. Satellite on-board processing of hyperspectral imaging data is desirable, but currently limited because large volumes of data need to be firstly downlinked for further processing, thus causing long lead times. To make the retrieval procedure more efficient, it would be ideal to have inversion algorithms capable of producing science data products on-board. Here, we describe how the Amenable Lookup Table Algorithm (ALTA; Gabrieli et al. 2017), which is a fast and compact Partial Least Square Regression (PLSR)-based technique for ground-based atmospheric trace gas retrievals, was modified to be employed for data processing on space-borne sensors. We refer to this new approach as the Adaptive Inversion Method (AIM). Here, we describe the new inversion algorithm and present preliminary results. We tested AIM on retrieving Land Surface Temperature (LST) from Hyperspectral Thermal Emission Spectrometer (HyTES) scenes and volcanic sulfur dioxide (SO2) from spectral imaging data of Kīlauea volcano, in Hawai`i, obtained using the MODIS-ASTER Airborne Simulator (MASTER). Results are encouraging and AIM may be suitable for being employed for data processing on-board future cube-sat missions with 10-100 spectral channels.

The land surface temperature (LST) CCI project aims to deliver a significant improvement on the capability of current satellite LST data records to meet the strict GCOS requirements for climate applications of LST data. Accurate knowledge of land surface temperature (LST) plays a key role in describing the physics of land-surface processes at regional and global scales as they combine information on both the surface-atmosphere interactions and energy fluxes within the Earth Climate System. This provides important information across a range of disciplines including monitoring drought, impact on human health, and changes in vegetation. Phase 1 of the programme of work has achieved some excellent progress: Detailed climate user input into the specifications of the LST ECV products, and user assessment of these products to drive LST exploitation in climate science Strong buy-in from the climate science community coordinated by the Climate Research Group A suite of high quality IR LST ECV Products and MW LST ECV Products for geostationary (GEO) and low earth orbit (LEO) satellites from the ATSRs, MODIS, SLSTR, SEVIRI, GOES, MTSAT and SSM/I An improved long-term LST CDR of +20 years from 1995 to 2020 for ATSR-2 through to SLSTR A +10 year Merged LST product combining the advantages of both GEO and LEO satellites Algorithm, cloud masking and uncertainty consistency across datasets We present here the approaches taken and the results to realise the full potential of long-term LST data for climate science.

Satellite-based estimations of land surface temperature (LST) are a valuable asset in the assessment of energy and water transfers at the Earth’s land-atmosphere interface. LST is most commonly estimated from radiometric measurements in the thermal infrared (TIR) atmospheric window (8–13 µm) using retrieval algorithms that account for land surface emissivity and atmospheric effects. However, current operational LST retrieval algorithms do not account for the effect of heavy aerosol loading on the retrievals. Here, we analyze the impact of high dust aerosol concentrations on three distinct LST products: (i) EUMETSAT LSA SAF’s SEVIRI product, which uses a Generalized Split-Window (GSW) algorithm; (ii) NASA’s MODIS product, MxD11, employing a similar GSW algorithm; (iii) NASA’s MODIS product, MxD21, which makes use of a Temperature-Emissivity Separation algorithm. We also perform radiative transfer simulations with RTTOV to study the radiative effects of heavy dust aerosol loadings on thermal infrared retrievals. The three LST products are first compared against ERA5’s skin temperature (SKT) across the Saharan Desert, where frequent seasonal dust production and transport occurs. Large anomalous differences are found between satellite LST and reanalysis SKT during summer months, coinciding with the highest dust aerosol optical depths at 550 nm from CAMS’ atmospheric composition reanalysis, EAC4. The LST products are also compared against in situ measurements at two ground stations in the Sahel region, showing increased biases for higher dust aerosol concentrations. Both comparisons – against reanalysis and in situ measurements – indicate that the three products analyzed underestimate LST in conditions of heavy dust aerosol loading. Analysis of brightness temperatures (BT) from the SEVIRI channels centered on 10.8 µm and 12.0 µm (used in LSA SAF’s GSW algorithm) reveals that dust aerosols have an opposite effect on BT differences compared to water vapor, which will introduce errors in the atmospheric correction if not properly accounted for. Preliminary radiative transfer simulations with RTTOV confirm this behavior of BT differences with dust aerosols and provide important information for addressing the effect of high concentrations of aerosols on thermal infrared LST retrievals. This work was performed within the framework of LSA SAF, with the aim of improving current LST retrieval methods.

Land Surface Emissivity (LSE) is a critical variable in the quantification of the surface energy budget and for the estimation of surface parameters from earth observation data, including Land Surface Temperature (LST). A widely used semi-empirical method to estimate LSE is the Vegetation Cover Method (VCM), however it originates large LSE uncertainties over desert and sparsely vegetated regions, due to the limited number of land cover types used to describe them. The TES algorithm, which is also extensively used, allows direct separation between emissivity and temperature and has been used to operationally produce LST and LSE based on multiple sensors. This method has shown to provide LSE estimates with good accuracy, however most validation exercises are conducted over desert sites. Validation over vegetated scenes is more complex given the high heterogeneity of surface elements with contrasting spectral characteristics. Over such surfaces, several authors have argued that emissivity estimates that use visible and near-infrared observations are generally amongst the most accurate, since the reduced spectral contrast decreases the accuracy of the direct retirevals. Furthermore, direct methods require accurate atmospheric corrections, being very sensitive to errors in the atmospheric data. A new LSE product is proposed that is based on the merge of the two widely used methods. Our aim is to take the best of each method, considering their differential performance over a wide diversity of surface conditions. As such, over vegetated areas, where spectral contrasts are reduced and retrievals using TES are more difficult, we use the VCM method, while over bare areas, where the VCM cannot estimate the spatial variability of the LSE, the TES is preferred. Furthermore, we propose a new calibration of the TES algorithm that allows a direct retrieval of the angular dependence of LSE. The new calibration makes use of the so-called multi-sensor method, where overlapping observations from different sensors are used to estimate the directionality of LSE. The proposed methodology was applied to observations from SEVIRI onboard MSG satellites and VIIRS onboard Suomi-NPP, to derive channel and broad-band emissivities in the 3-14 µm range. The product shows good agreement with in-situ, with accuracies of 0.009 and 0.014 in the 8-14 µm and 3-8 µm regions, respectively. The methodology described in this article will be used by the LSA-SAF for LST production from current and upcoming EUMETSAT missions.

Here we present a comprehensive database of atmospheric profiles and surface variables of relevance for Land Surface Temperature (LST) models using Thermal Infrared (TIR) observations. The database was built from the European Center for Medium Range Forecast (ECMWF) version-5 reanalysis (ERA5) dataset. Reanalysis data is particularly useful to build a calibration database since it combines large amounts of historical observations with the most advanced modeling and data assimilation systems. Moreover, they provide a large set of surface and profile variables that are consistent with each other and are available at full spatial and temporal coverage. This calibration database is built by sampling atmospheric profiles of specific humidity and temperature from the ERA5 dataset, using a dissimilarity criterion developed by Chevallier et al. (2000) for the TIGR databases. Other ERA5 variables corresponding to the selected profiles that are relevant to the LST are also included in the database, namely profiles of ozone, 2-meter temperature (t2m), surface pressure, skin temperature (Tskin), total column water vapour (TCWV) and total cloud cover (TCC). Despite the great advances in surface modelling in the last decades, modelled Tskin still have significant errors. Tskin estimates should be used carefully in the context of algorithm calibration, since the errors, in particular the systematic ones leading to undersampling of the Tskin actual distribution, will be propagated to the calibration process and could compromise the quality of the algorithm. To reduce the impact of such errors on the database, we complement ERA5 surface information with LST and emissivity estimates from multiple satellite products. Our strategy is to define an acceptable range of values of LST given the atmospheric conditions, thus increasing the representativeness of the database. Similarly, for the emissivity we take realistic ranges of values based on satellite products obtained for each land cover type. This work was carried out within the framework of the Satellite Application Facility on Land Surface Analysis (LSA-SAF) with the purpose of creating a training database for the development of LST retrieval algorithms for the next generation of satellites from the European Organization for the Exploitation of Meteorological Satellites (EUMETSAT), the Metop Second Generation and the Meteosat Third Generation. We will also show some applications of the dataset to the development of LST retrieval algorithms in the context of the LSA-SAF.

Thermal images are widely used for a range of downstream tasks such as forest fires, volcanology, military applications, soil moisture studies, hydrology, and coastal zones. Thermal images contain thermal emission of the observed object and, therefore, are dependent on the type of the object. Considering that different land cover types have different thermal emissions, the goal of this study is to retrieve the land cover type information from a single thermal image in the absence of the cloud. We aim to find out which land covers can be extracted from the Landsat thermal images using state-of-the-art machine learning techniques. For this purpose, we created a dataset containing geographically well-distributed 8665 Landsat thermal band 10 images with 100 meter ground resolution. The images are in the size of 512 by 512 pixels and their cloud coverage is less than 5%. The ground truth land cover label of each image comes from ESA Worldcover classes for the corresponding area with similar size and resolution. The initial investigation is conducted on the water classes of the images, where we train a UNet to only detect the water bodies and test on samples from the test set. The results after 40 epochs of training are promising, as the model was able to detect the main rivers and the sea areas. The number of false positives in the test images with no water pixels is considerably low. The next step would be extracting each of the single classes and then combining them. To sum it up, we explore which of the land cover classes can be retrieved from a single thermal image. We also look for the best solution among the semantic segmentation methods to classify the land covers. Our initial experiment on detecting water bodies shows that thermal images indeed have a different sensor output to water rather than other classes on earth. Next, we seek the best solution to extract each of the single classes and the combination of them from the thermal Landsat band 10 images.

The continuous growth of infrared-based remote sensing applications in recent years has led to an increasing demand for high spatial resolution thermal infrared images, e.g. for the monitoring of urban heat islands, irrigation management and wildfire detection. The native GSD of available satellite instruments is often not sufficient for specific use cases. Software-based techniques to increase resolution, in particular deep-learning based super-resolution techniques, have attracted much attention in recent years to improve the quality of low-resolution remote sensing images, especially in the visible domain. In this work, we discuss the challenges of transferring established super-resolution algorithms from the visible to the thermal infrared spectrum. The techniques discussed encompass both, single- and multi-image as well as single- and multi-band methods. We carefully evaluate the models’ ability to adaptively reconstruct higher resolution details, as many of these techniques have been developed with applications for consumer technology in mind and can introduce artifacts. Thus, we evaluate the different approaches with regards to their radiometric consistency, artifact introduction and uncertainty quantification. We use Landsat-8’s Thermal Infrared Sensor (TIRS) Level 2 data as well as ASTER Level 1B data as reference datasets, both providing long-wave infrared (LWIR) data at a GSD of 90-100m. We evaluate the different models against the baseline of bicubic upsampling using PSNR, SSIM and LPIPS metrics to compare their performance considering signal strength, structural and perceptual similarities in different land cover classes. However, we show that these algorithms are capable of making use of physical information available through time series or auxiliary bands.   Furthermore, we discuss possible applications of super-resolved datasets, their limitations as well as future research directions, based on the challenges we identified.

In the next decade(s) a set of satellite missions carrying Thermal InfraRed (TIR) imagers with relatively high NEdT is foreseen, e.g. the high resolution TIR imagers flying on the future TRISHNA, LSTM, SBG missions or the secondary payload on board of the ESA Earth Explorer 10 Harmony. The instruments on board these missions are expected to be characterized by NEdT of the order of tenths of degree K. In order to reduce the impact of radiometric noise on the retrieved SST, this study investigates the possibility to apply a multipixel atmospheric correction (Harris and Saunders 1997, Merchant et al. 2013) based on the hypotheses that: i) the spatial variability scales of radiatively active atmospheric variables is, on average, larger than the one of SST; ii) the atmospheric correction is based on the split window difference. Based on a set of SLSTR cases covering different regions, in the global oceans, characterized by high spatial variability of the SST, the study demonstrates that the local spatial variability of the split window difference term on scales of ≃ 3x3 km, is comparable with the noise associated to the measurements. Similarly, the power spectra of the split window term is shown to have, for small scales the behaviour of a white noise spectra. On this basis we suggest to average, on a proper scale that can be dynamically defined for each pixel, the split window term and to use the average for atmospheric correction reducing the impact of radiometric noise.

As for all optical earth observation missions, one of the essential steps of the pre-processing phases includes the detection of clouds and their shadows, as well as the correction of atmospheric effects. For more than 15 years, CNES and CESBIO have been developing a processor named MAJA for the cloud detection and the correction of atmospheric effects. Its particularity is the use of multi-temporal information to improve the cloud detection and the atmospheric correction. MAJA has been intensively validated and is now used in many processing centres, to process Sentinel-2 and VENµS data, within CNES for the Theia land data centre, within the DLR, within the Copernicus Snow and Ice processing centre, at the National Mapping Agency of Norway, or within the Sen2AGri and Sen4PAC projects. MAJA is an open source software already downloaded about 2500 times. The case of TRISHNA introduces new opportunities, but also new challenges for its atmospheric correction. The availability of thermal infrared bands can improve the detection of clouds, but the very wide field of view (viewing zenith angle may reach 40°) is clearly a difficulty. TRISHNA's orbit has an 8 day repeat cycle, but a 3 day sub-cycle. It means that a given pixel can be observed at least 3 time every 8 days. The viewing angles are identical 8 days apart, but the differ within the 8 days cycle.. MAJA uses a multi-temporal method that compares two successive cloud free acquisitions to detect clouds and estimate aerosols. The differences in surface reflectances due to directional effects with different viewing angles could degrade the estimates. We compared two options in the processing : - using MAJA considering only the previous acquisitions obtained with the same angles every 8 days, - using MAJA with all the viewing directions but with a directional effect corrections. The first option degrades the temporal revisit considered within MAJA, while the second one may still be sensitive to the residuals of the directional effect correction. To have an idea of the best solution, we tried both approaches using OLCI data which also has a large field of view. The results showed that the results stay correct for both approaches, even if the cloud detection works a bit better with the second option, while the atmospheric correction is better with the first one. New improvements can take advantage of both options.

In the context of the TRISHNA mission preparation, we have conducted a thorough analysis of the Temperature-Emissivity Separation (TES) method and its related literature in order to study the relevance of its use during the operational phase of the mission. This analysis led us to propose a more mathematical approach to the TES method by considering the εmin /MMD relationship as the additional equation necessary to solve the ill-posed problem of emissivity/temperature separation. With such a deterministic system, emissivity and temperature can be obtained using a classical optimization approach with an initial condition. We considered such an approach by introducing a new spectral invariant in order to test convergence of the process. This new formulation is tested against the original version of TES over a wide range of realistic scenarii including vegetation canopy-scale cavity effects and realistic instrumental noise. Despite a small gain in performance on the estimation of surface temperature with a difference in RMSE of 0.03 K, the spectral invariant based TES approach (SITES) shows numerous advantages as compared to the original TES approach. First, it removes the ambiguity on the convergence test that exists in the original TES by testing convergence on a single parameter as opposed to 4 radiances values, one per TRISHNA channels. Second, the SITES approach shows a significant increase in emissivity estimation performance, with differences in RMSE of 1.3, 1.2, 1.7, and 2.6, for TIR1, TIR2, TIR3 and TIR4, respectively. Third, the SITES method appears to converge faster than the original TES, with a maximum and average number of iterations of 6 and 2.07 respectively, as compared to 12 and 2.13 for the original TES, disregarding the possible iterations made for the εmin refinement. The SITES demonstrates overall better performances than the original TES approach, and therefore appears as a better candidate for use during TRISHNA operational phase.

Estimating the Land Surface Temperature (LST) from remotely sensed thermal infrared data is only possible under clear-sky conditions. To tackle this problem, recent efforts investigated the fusion with passive microwave measurements and the use of land surface energy balance models to fill the cloud gaps, resulting in the first generation of seamless all-weather LST products. These products, however, continue to suffer from the trade-off between the spatial and temporal resolution and as such cannot provide LST data with high spatial and temporal detail. In this work we present a method for addressing this limitation by downscaling all-weather LST with high temporal resolution. The proposed method uses a Random Forest (RF) regressor with an extensive set of LST predictors that describe the land cover, the topography, the vegetation, the satellite viewing geometry, and the cloud spatial distribution. In contrast to traditional approaches that directly downscale the LST, the RF regressor is trained to predict the LST residuals that are calculated as the difference between all-weather LST and corresponding modelled clear-sky LST. The modelled data are derived from the LST Cycle Parameters (CP) presented in Sismanidis et al. (2018, 2021) that provide a seamless pixel-based climatology of the annual and diurnal dynamics of clear-sky LST and are available at the target (fine) and source (coarse) spatial resolutions. The RF regressor is trained with the coarse resolution residuals and then used to predict the fine resolution residuals. The downscaled all-weather LST are then produced by adding the predicted residuals to the corresponding modelled clear-sky LST generated using the fine resolution CP. The proposed method is tested over mainland Europe using four months (July-October 2021) of diurnal half-hourly all-weather LST obtained from EUMETSAT’s Satellite Application Facility on Land Surface Analysis (Martins et al. 2019). The source and target resolutions are 0.05 deg and 0.01 deg, respectively, and the results are evaluated using independent satellite and in-situ LST data from the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Evora station in Portugal.

Since the advent of data science, time series analysis has been employed to forecast vegetation dynamics and identify future patterns and trends along with monitoring and detecting land cover changes. A wide range of models is utilized regarding time series forecasting, which includes statistical methods such as automatic regression models and others based on more sophisticated machine learning methodologies such as Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Recurrent Neural Networks (RNN), and Long Short Term Memory Network (LSTM). In this study, we used vegetation time series to train a neural network model called Conv-LSTM, which combines CNN model and LSTM model to predict future patterns throughout the course of the next ten years (2023-2032). Using MODIS/Terra sensor, we collected satellite images of the Tanger-Tétouan-Al Hocema (TTA) region of Morocco from 2010 to 2022, and we selected four images to represent the four seasons in each year (i.e., winter, spring, summer, and autumn). Then, we computed 10 vegetation biophysical indices [i.e., Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), Enhanced Vegetation Index (EVI, EVI2), Global Environmental Monitoring Index (GEMI), Difference Vegetation Index (DVI), Transformed Vegetation Index (TVI), Renormalized Difference Vegetation Index (RDVI), Green Ratio Vegetation Index (GRVI), and Plant Senescence Reflectance Index (PSRI)] using self-developed software. Lastly, we used the Conv-LSTM model to forecast vegetation trends from 2023 to 2032. The study revealed that the average values of GNDVI, NDVI, and TVI are expected to decrease by 10.6%, 2.8%, and 0.3%, respectively, by 2032 during the peak season. On the other hand, it is anticipated that the average values for EVI, EVI2, RDVI, GRVI, DVI, PSRI, and GEMI will rise by 20.1%, 16.8%, 15.7%, 6.4%, 6.2%, 2.7%, and 1.1%, respectively. Regarding the low season, it is estimated that the average values of GNDVI, PSRI, NDVI, and TVI will increase by 11%, 6%, 4.5%, and 0.1%, respectively. Conversely, DVI, RDVI, and EVI2 will all decrease by 12% over the next ten years.

The Land Surface Temperature (LST) is sensitive to the energy and water exchanges at the land-atmosphere interface and hence, is extensively used in a variety of applications related to hydrology, water resources, vegetation monitoring, agriculture, urban studies, weather, and climate modelling. Observation from Thermal Infrared (TIR) sensors aboard multiple satellites enables LST retrieval at fine to coarse spatial resolutions (100 m to 5000 m) with an accuracy of 1–2 K. However, the radiation emitted by the earth in the TIR band is incapable of penetrating clouds resulting in long gaps in LST time series which sometimes span several months. On average, 60% of the land surface is covered by clouds, limiting the data availability from TIR sensors. This severely hinders the applications that depend on LST data. Apart from TIR bands, the radiation from the earth can also be observed in microwave (MW) wavelengths using passive microwave radiometers. MW radiation can penetrate clouds, potentially providing all-weather surface information albeit with a much coarser spatial resolution (~10 km to 60 km). In this research, we employed a Random Forest (RF) algorithm to examine the association between PMW Brightness temperatures and TIR LST on a 1km scale over the Indian geographical region. Since the LST is closely related to land cover, location, Day of Year, terrain, and vegetation conditions, these variables were selected as additional inputs to improve the accuracy of the RF model. When compared with the MODIS LST, the model shows an average RMSE of 3 K during daytime and 1.9 K during nighttime, with the coefficient of determination (R-Square) of 0.89 and 0.91, respectively. Validation of the predicted LST using in situ observations are underway and further, the model's effectiveness will be evaluated under varying conditions including latitude, elevation, and vegetation cover to understand the model performance thoroughly.

There have been little efforts to the angular variation of remotely sensed surface temperature for simplified urban neighbourhoods with physically-based and parametric models, but research is at early stage and far from being operationally applied with actual satellite data of urban areas. The urban surface has particular properties, which affect the physical processes occurring in the urban canyons and hinder the estimation of urban surface temperatures from space. Geometric properties, including orientation and openness to sun and sky provide a strong control on urban surface temperature. Moreover, the emissivity of anthropogenic materials presents large variations in emissivity. If detailed information on the roofing, façade and paved surface materials found in a city is available, their corresponding emissivity can be approximated with higher confidence. Detailed surface cover maps, including buildings, façades and paved surfaces materials are used in this study for the emissivity estimation, using ancillary spectral library information to link material types with their representative emissivity values from the spectral library. ECOSTRESS images for the city of London are used for assessing detailed urban surface temperature. The fractional surface cover corresponding to the ECOSTRESS pixel, is estimated for every acquisition, to ensure the appropriate 3D urban surface cover depending on the viewing angle. Results are very promising with evaluation against the ECOSTRESS products revealing a mean absolute error of 1 K. Evaluation with in-situ measured urban temperatures from radiometers is in progress.

At the end of 2017, the National Academies of Science, Engineering, and Medicine, made recommendations to NASA and other agencies, by outlining the most pressing science concerns for the decade. This Decadal Survey (DS), titled “Thriving on our Changing Planet”, called for a mission to map the Earth’s Surface, Biology, and Geology (SBG) in order to answer science questions in the fields of ecology, hydrology, climate, and solid earth. In 2018, an SBG Algorithms Working Group (Alg WG) was formed, and this group has continued to meet regularly for almost four years. The Alg WG mailing list contains more than 250 people, and more than 40 people (with varying audience by topic) typically attend each biweekly telecon. The Alg WG is entirely open to all who would like to join, and includes scientists and industry representatives from around the world. The Alg WG has a formal charter to support mission concept development by assessing the status of existing algorithms, identifying gaps and opportunities, and assisting in traceability studies. With this in mind, the first working group activity was to gather a list of products that could be used to answer science questions across the fields of snow/ice, volcanoes, aquatic and terrestrial ecosystems, and minerals/soils. The combined list exceeded 200 products and associated algorithms, and about 100 of these are described in overview published in 2021. The second task of the Alg WG was to reduce this all-encompassing list to one that was achievable, judged by algorithm maturity and relevance to the DS. In this presentation we consider only the thermal infrared products. From acquired radiance data, land surface temperature and emissivity are considered “base products”. Beyond these base products, the WG defined a list of high-priority products. For the thermal IR, these included: substrate composition; volcanic gases and plumes; high-temperature features; and evapotranspiration. An important outcome of this work was the observation that a midwave infrared (MIR) band (not explicitly called out in the original architecture proposals) was required for a number of algorithms under consideration, and this finding has resulted in a modified proposed architecture: a multiband TIR instrument with early afternoon overpass, >5 TIR bands, >1 MIR band, 60 m GSD, and 3 day revisit. Low-latency products were also a high priority for the applications community, particularly relating to hazards (volcanic precursors, drought monitoring, etc.). In this presentation we will summarize the findings of this working group.

One of the top priorities of the Surface Biology and Geology (SBG) Earth Observing System (EOS) is the detection and modeling of extremely high-temperature phenomena (> 400 K), as it is critical for studying natural hazards such as active fires and volcanic eruptions. As a precursor to the mission, we test whether the current midwave (MIR) and thermal infrared (TIR) band specifications, including noise levels, saturation levels, and band position, are sufficient to be able to detect high temperatures and thermal anomalies. Specifically, our investigation aims to quantify the use of the 4 and 4.8 μm MIR bands for detecting and retrieving high-temperature features in the 400-1500 K range. We utilize the Land Surface Temperature data obtained by the airborne Hyperspectral Thermal Emission Spectrometer (HyTES) instrument over fire and lava locations. We use these to model the at-sensor SBG radiances using the spectral response functions and instrument noise model in MODTRAN for the designated/proposed MIR and TIR channels. For hotspot detection, we apply MODVOLC's Normalized Thermal Index (NTI) and MIROVA's Enhanced Thermal Index (ETI) to determine a suitable threshold. We find that an approach combining NTI threshold of -0.7 followed by an ETI threshold of 0.02 serves to identify hot anomalies with the highest detection accuracy of 97%. The effect of noise is only noticeable under 400 K, so it does not reduce the detection accuracy for hot anomalies in the 400-1500 K range.

With the unprecedented high resolution, frequent revisit time and long-term data availability promised by the next generation of thermal missions, Trishna, SGB and LSTM, new calibration/validation methodologies, new surface energy budget models and new applications are to be developed. This poster addresses modeling the surface temperature of snow-covered areas in mountainous regions. Indeed, large spatial variations of the surface temperature (>10 K) over small horizontal extents are commonly observed in mountains due to the extreme variety of slopes, altitudes, and orographic conditions. These variations will be better captured by the next generation missions thanks to their improved spatial resolution, which implies at the same time improving our understanding of the physical origin of these variations. Although modeling the surface energy budget for a flat surface with an infinite horizontal extent is a common task in meteorological and hydrological modeling, significant additional work is required to account for the modulation of the short-wave irradiance by the local slopes, the shadows, the reillumination between the surrounding slopes in the short-wave and in the long-wave, the influence of slope on the turbulent fluxes, the altitudinal atmospheric variations, the wind flow around the relief, the decameter-scale surface heterogeneity. This poster presents the ongoing development of a modeling chain to compute the surface temperature at decameter resolution in mountains. The first component of this chain is the Rough Surface Ray-Tracing (RSRT) model. Based on a photon transport Monte Carlo algorithm, this model calculates the incident and reflected short-wave radiation on every facet of the mesh describing the terrain. The second component is a surface scheme that estimates the energy fluxes between the surface and atmosphere and deduces the surface temperature. An initial assessment at the Col du Lautaret, in the French Alps, shows an agreement between the simulations and local observations within 0.2∘C in winter, and a satisfying high spatial correlation with Landsat 8 and 9 observations. The direct effect of short-wave modulation by the slope is found to be the main driver of these variations, during clear-sky days. Now that the surface radiative scheme is improved, the next steps includes improving the variable long-wave contribution of the atmosphere, the heterogeneity of the surface (snow/grass/soil), and spatial variations in the wind flow. Beyond the calibration/validation of thermal sensors, this modeling chain will be useful to better estimate snow melt for hydrological applications, ground temperature for ecological applications, and surface-atmosphere fluxes for micro-meteorological applications.

Associated authors: S. Dransfeld (2), V. Levasseur (1), J. Bruniquel (1), J-L Roujean (3), P. Gamet (3), J-P Gastellu-Etchegorry (4), B. Pflug (5), RD. Delosreyes (5), D. Ghent (6), K. Mallick (7), D. Smith (8), J. Fischer (9), R. Preusker (9), J. Sobrino (10), J. Jackson (11) (1) ACRI-ST, 260 Rte du Pin Montard, 06904 Sophia-Antipolis - France (2) ESA/ESRIN, Via Galileo Galilei, 1, 00044 Frascati RM, Italy (3) CNES, 18 Av. Edouard Belin, 31400 Toulouse, France (4) CESBIO, Centre d'Etudes Spatiales de la Biosphère, 31400 Toulouse, France (5) DLR, Pfaffenwaldring 38-40, 70569 Stuttgart, Germany (6) University of Leicester, University Rd, Leicester LE1 7RH, UK (7) LIST, 5 Av. des Hauts-Fourneaux, 4362 Esch-sur-Alzette, Luxembourg (8) RAL, Fermi Ave, Harwell, Didcot OX11 0QX, UK (9) Spectral Earth, Baseler Str. 91A, 12205 Berlin, Germany (10) University of Valencia, Av. de Blasco Ibáñez, 13, 46010 València, Valencia, Spain (11) ARGANS, Science Park, 1 Davy Rd, Plymouth PL6 8BX, UK ___ Abstract: The purpose of the presentation is to introduce the content of the LSTM-L2 project beginning Q2 2023 with a description of the consortium, planning, content of the products and objectives. The purpose of the project is to develop and to deliver the L2A operational processor which will be deployed in the LSTM ground segment and be used by the Agency to generate the LSTM L2A products. The L2 operational products to develop refer to Land Surface Temperature and Land Surface Emissivity retrieval, Water Vapor computation, Aerosol Optical Thickness retrieval, Atmospheric correction, and Cloud masking. Within the framework of the project, a certain number of activities will be conducted, in particular: ❖ Select and define the algorithms used for the retrieval of the various parameters (LST, LSE, aerosols, TCWV…) and to improve these algorithms over the whole duration of the project, ❖ Develop a prototype used as a precursor and to test any evolution of the algorithms prior to their implementation into the operational processor, ❖ Define the cal/val activities, before the launch, during the phase E1 (commissioning) and during the Phase E2 (routine operations), needed to assess the performances and the quality of the operational LSTM L2A products. ❖ Support the engagement of a community of users and facilitate their use of LSTM products, develop open-source library modules allowing them to process the products themselves (the prototype will also be made available to users).

Previous global L4 sea surface temperature (SST) analysis inter-comparison studies were centered on the assessment of the accuracy and bias in the various L4 SST by comparing them with independent near-surface Argo profile temperature data to assess their consistency. This type of assessment is centered in the absolute value of SST rather than in the SST differences (gradients), which is more relevant to the study of oceanographic features (e.g., fronts, gradients, eddies, etc) and ocean dynamics. Here, we use for the first time a metric, the spectrum of singularity exponents, to assess the structural and statistical quality of different L4 GHRSST products based on the multifractal theory of turbulence. The singularity exponents represent the geometrical projection of the turbulence cascade, and its singular spectrum can be seen, roughly, as the probability density function (PDF) of the singularity exponents normalized by the scale. Our results reveal that the different schemes used to produce the L4 SST products may contribute to the loss of dynamical information or structural coherence. This new metric constitutes a valuable tool to assess the structural quality of SST products and can support data satellite SST producers efforts to improve the interpolation schemes used to generate L4 SST products.

Hydrosat’s prototype longwave infrared imaging (LIRI) system consists of two bands, centered at approximately 10.9 µm and 12 µm, with expected ground sample distance of 70 meters. An 8-band visible to near-infrared imager (VIRI) will collect coincident reflective data. After launch, onboard calibration will be performed to update pre-launch calibration. To validate the absolute radiometric calibration, vicarious and cross-calibration techniques will be applied. The vicarious calibration will involve using the National Oceanic and Atmospheric Administration (NOAA) ocean and great lakes buoy temperature data. Due to its high known emissivity, water has long been used for remotely sensed thermal calibration. Measured buoy subsurface temperatures will be adjusted to water skin temperature and modeled to sensor reaching radiance using local weather data and radiative transfer modelling. As a secondary validation method, cross calibration with other spatially and temporally coincident thermal sensors will be used to monitor changes over time. The calibrated top-of-atmosphere radiance data will then be ingested into the generalized split window algorithm to create a preliminary land-surface-temperature (LST) product. Validation of the LST product will include the above-mentioned buoy data, as well as the SURFRAD (surface radiation budget) network sites. These sites span a range of climatologically diverse regions and measure up- and down-welling broadband thermal irradiance every few seconds. The calibration and validation strategy are instrumental to assist with Hydrosat’s goal: daily high-resolution land surface temperature to help manage Earth’s most valuable resource: water.

We present new methods for physical interpretation and mathematical treatment of the imaging contrast observed in thermal infrared (TIR) images of the rocky upper scarp of the Poggio Baldi landslide (Italy), which is part of a natural laboratory. Exemplar thermal images have been acquired with a high-performance camera at a distance around 500 meters, in a geometry where reflection is expected to dominate over thermal emission. The digital pixel intensities have therefore been considered as wavelength-integrated infrared spectral reflectance, irrespective of the temperature scale loaded into the camera software. Sub-portions of the scarp producing lower signal have been identified by a multiscale image segmentation algorithm and overlaid on the visible image to provide an interpretation for the different thermal imaging contrast mechanisms that may be exploited for landslide monitoring in the future. We have found that the TIR image contrast analysis can highlight different physical mechansims behind TIR contrast: (i) a different local orientation of the rocky wall if compared to the average scarp surface orientation, due to the fact that reflectance will in general dominate over emittance of rocky walls, but more so for properly oriented scarp sub-portions; (ii) a different grade of humidity of scarp sub-portions, because, according to our physical model of the optical constants in the TIR wavelength range, even a surface water layer of 100 micrometer thickness can decrease the TIR signal to an extent that can be observed by a high performance TIR camera; (iii) a different mineral content in scarp sub-portions, due to the high sensitivity of TIR spectral reflectance to the specific mineral oxidation state, which is related to the exposure time to weathering agents.

Land Surface Temperature (LST) is an essential climate variable (ECV) which yields critical information about the Earth’s radiative energy budget and helps to constrain climate models, as well as providing information about temperature changes in remote regions. Satellite LST datasets are required to have a spatial resolution of < 1 km and a measurement uncertainty of < 1 K to meet WMO Global Climate Observing System (GCOS) requirements. The Sea and Land Surface Temperature Radiometer (SLSTR) aboard the Sentinel-3 satellites A (launched in 2016) and B (launched in 2018) are capable of producing such datasets, but require rigorous ground-based validation to confirm this. Absolute validation of satellite LST is only possible via comparisons with in-situ observations of thermal radiation. A well-established suite of measurement sites exist as part of long-term monitoring networks (e.g. ARM, SURFRAD), and are routinely used for validation of SLSTR within the Sentinel-3 Mission Performance Centre (S3MPC) and will be for its successor the Optical Mission Performance Cluster (OPT-MPC). These sites do not however fully account for all possible biomes on the Earth’s surface, as mosaic vegetation and broadleaf deciduous forests are not represented by existing measurement sites. Therefore, to increase the scope of Sentinel-3 validation requires the deployment of new measurement sites, in addition to the comparisons with existing monitoring networks. The “Copernicus Space Component Validation for Land Surface Temperature, Aerosol Optical Depth and Water Vapour Sentinel-3 Products” (LAW) project performed an extensive and systematic validation of Sentinel-3 datasets against ground-based observations through calibration, instrument deployment and subsequent validation matchups for 5 new LST observation sites in previously unobserved biomes: KIT forest (Germany): closed broadleaved deciduous forest Svartberget (Sweden): open needleleaved deciduous or evergreen forest Hyytiälä (Finland): closed to open mixed broadleaved and needle leaved forest Robson Creek (Australia): closed to open (more than 15 %) broadleaved evergreen and/or semi-deciduous forest Puéchabon (France): sparse vegetation This presentation will cover the progress made in deploying these new stations, comparisons between Sentinel-3 and ground-based data, and potential consequences for refining the Sentinel-3 LST retrieval algorithm based on these analyses.

In field validation sites are category ‘A’ validation sites for satellite measurements, otherwise said they are the reference for validating surface temperature. Ideal sites are few and far between and they mainly situated over water bodies. This is of course related to the thermal stability of water bodies, where emissivity is homogeneous and well characterised thus removing major uncertainties in surface temperature measurements. Furthermore, the effect of turbulence over waters is extremely small, which further permits precise surface temperature estimates. However over typical land surfaces and in particular vegetated surfaces, which are of interest to the scientific community in the context of thermal missions given their potential role in the detection of water stress, it remains important to validate satellite measurements of vegetated surfaces. Perhaps not for absolute calibration as such sites will always be more “noisy” than water bodies but at least for high quality validation of land surface temperature. In addition, in the context of the TRISHNA, SBG and LSTM missions, with their improved thermal and temporal resolution, it will become both important and easier to find sites homogenous at the scale of a few satellite pixels. It is therefore essential to optimise our land validation protocols with these missions in mind. One important problem remains the turbulence that can change surface temperature by over a few degrees in a relatively short time scale making surface temperature estimates “noisy”. How can we improve our estimates or at least quantify errors ? This poster will revisit in-situ field site measurements and develop a protocol to provide a better estimate of surface temperature.

KOMPSAT-3A is a Korean polar-orbitting satellite that hosts a thermal sensor for the first time in the KOMPSAT (Korea Multi-Purpose Satellite) series [1]. Since its launch in 2015, the satellite has produced thermal images of mid-wave infrared in high spatial resolution (5.5 m) and high image quality [1]. Even though the satellite does not have an on-board calibrator, the MWIR band has not been vicariously calibrated so far, thus failing to produce quantitative temperature retrieval for both ground and ocean. VC of the sensor is essential for accurate estimation of surface temperature as well as for monitoring long-term sensor stability [2]. The VC results of Landsat-8 Thermal Infrared Sensor (TIRS)  showed that the sensor had an apparent calibration error (-2.1 K and -4.4 K, respectively for Band 10 and Band 11) [3]. The 5 thermal infrared bands in Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) were vicariously calibrated not only for water targets but also for land sites, which exhibited relative difference of 0.5 % or better over the radiance range  6.5–13 W/m2/sr/μm [4]. In this study, we conducted vicarious calibration for the MWIR band of KOMPSAT-3A, based on buoy data coincident with satellite observation. Firstly, atmospheric profile and surface temperature data were collected for available past images, and radiative transfer simulation was run using MODTRAN v.6. Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2) data were used for retrieving atmospheric parameters such relative humidity and atmosphere pressure [5]. Water temperature at the field was obtained from an archive of National Data Buoy Center (NDBC) [6].  Finally, top-of-atmosphere radiance was simulated for several buoys of various seasons to derive the statistical vicarious gain and offset of the thermal band.

The aim of this report is to document the spatial distribution of Land Surface Temperatures (LST) in different regions of the Continental US (CONUS), and the Meso-America (Central America, the Caribbean, and Northern Regions of South America) using NOAA GOES-16 datasets high temporal resolution. The most significant difference from the operational LST product is the frequency of temporal sampling, which is once every 5 minutes instead of daily consequently increasing the quality of the end result. The LST accuracy standard for all ABI scanning modes in the GOES-R program is 2.5 K. (i.e., full disk, CONUS, and mesoscale). Our work considered data sets between Jan 2017 and December 2021, 5 minutes resolution for both CONUS and Mesoamerica. Climatology of maximum and minimum temperatures compares very well with NASA LANDSAT for the same period. Hourly climatology is also compared with the 5 minutes data sets showing minimum differences. The monthly winter maximum LST spatial distribution illustrated the highest values in Mexico and Greater Antilles within the range of 300 to 320 K. These are the same regions to have experience relatively higher minimum LST on average. The maximum daily temperate range was found to occur in the month of June exhibiting large range in the southwest of the CONUS (60⁰F) and minimum in Central America (20⁰F). The highest maximums were found in the month of August ranging from 320 K to 340⁰F while the June minimum LST values vary from 290K to 300K, in comparison. The annual maximum temperatures (TmaxJuly-TmaxJan) was found to range between 10-60⁰F. Diurnal cycles for selected urban sites (New York City, Chicago, Los Angeles, Mexico City) are also shown for winter/summer, showing expected patterns between Northeast, Southwest, Great Lakes regions, and high elevation Central America, demonstrating the added value of GOES-16 temporal resolution for local studies.  

Thermophysical remote sensing data are commonly utilized to measure the composition and size-frequency distribution of rock fragments excavated during impact crater formation (i.e., the ejecta deposit) on terrestrial planetary bodies. Such measurements have the potential to improve our understanding of the age, geologic history, and environmental conditions associated with the surface into which the impact crater has formed. However, orbital-based thermophysical data commonly lack the resolution necessary to resolve small-scale features that could enhance our understanding of the mechanics and environmental effects of impact crater formation. The objective of the work presented here is to define geologic unit boundaries and rock fragment size distribution within the Barringer Meteorite Impact Crater (“Meteor Crater”) ejecta deposit using drone-based and orbital thermophysical data. We use surface temperature data from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) (90 m/px) and from a FLIR (Forward Looking Infrared) Vue Pro 640 R thermal camera with a 9 mm lens (~0.25 m/px) attached to a DJI Phantom 4 Pro sUAS (small Unmanned Aircraft System, aka “drone”) to derive the Apparent Thermal Inertia (ATI) of the Meteor Crater ejecta deposit. The ATI measurements derived here are sensitive to the surface rock fragment distribution and degree of induration associated with the ejecta. Results indicate that ejecta distribution patterns are well behaved at the scale of the ASTER ATI data, but significant variability exists within the higher-resolution drone-based ATI data. The inconsistencies between ASTER and drone-based ATI values may be a result of local, human-induced erosion within the ejecta deposit, error associated with the predefined geologic unit boundaries used to bin our ATI data, or the result of ejecta distribution processes that are more complex than previously expected. Such scale-dependent factors should be considered when mapping and analyzing ejecta deposits on Earth and beyond.

The Mediterranean regions are particularly threatened by flash floods. They represent one of the greatest natural hazards in the High Atlas. Therefore, mastering the mechanisms of their occurrence in small mountainous watersheds is an important aspect of these difficult-to-control hazards. Flash floods are usually extreme hydrological events that are poorly recorded in regions where there are few and spatially poorly distributed monitoring stations and are characterized by high seasonal variability of hydrometeorological data. Considering the great influence of the uncertainties of the calibration parameters in hydrological models, it is difficult to predict their behavior, as in our case for the Zat River basin. The aim of this study is to understand the seasonal behavior of runoff, rainfall, and surface temperature in the Zat River basin by calculating the model uncertainty using the sensitivity parameters. The analysis was developed using instantaneous rainfall, runoff, and surface temperature data on the 10-minute time scale during the period from 01/09/2011 to 31/08/2018. More than 100 flood events were simulated and calibrated with the HEC-HMS model. A sensitivity parameter calculation approach was implemented, where three sensitivity parameters were identified, namely: curve number ‘CN’, concentration-time ‘TC’, and peak discharge "QM". To analyze the uncertainty of the calibration parameters, the probability distribution function and Monte Carlo simulations were applied to analyze the uncertainty of the curve number, time of concentration, and peak flow. The temperature index approach applied in hydrological modeling indicates that the snow water equivalent is the main source of uncertainty in the model, as it is directly influenced by temperature and therefore influences the discharges. The results showed that the observed and simulated hydrographs were highly correlated. In addition, the model performance was evaluated with a Nash coefficient ranging from 61.9% to 90% at the calibration level. This study is considered one of the first approaches to calculating the uncertainty in this region. Therefore, the established approach could be developed in other regions to improve flood forecasting and disaster management.

A regular laboratory calibration of Thermal Infrared (TIR) field radiometers is often infeasible, because they are installed at remote sites on autonomous ground stations. The next best approach to ensure the correct performance of field radiometers are in-situ calibrations. However, these are frequently complicated by difficulties to access the stations, travelling restrictions or simply cost and limited resources (e.g. qualified staff). Therefore, we developed two tests for continuously monitoring the stable performance of downward-looking TIR field radiometers; the only condition for applying the tests is that additional air temperature (AT) measurements are available. The first test (Test 1) is based on the difference between ground brightness temperature (BT) and AT, which is generally small and constant under a homogeneous sky, i.e. clear sky nights and days and nights with a homogeneous and full cloud-cover. The second test (Test 2) assumes that, under clear sky conditions and in the absence of advection, BT and AT are twice per day equal, i.e. ‘cross’ each other: this takes place near sunrise and sunset, here termed ‘morning/night crossing time’. Under such conditions the temporal difference between sunrise (sunset) and the morning (night) crossing time of BT and AT is expected to be relatively constant over the year with only slight seasonal changes. Both tests were applied to the five LST validation stations of the Copernicus LAW project (https://law.acri-st.fr/home), which are located in forests at different latitudes, i.e. in Karlsruhe – KIT (Germany), Hyytiälä (Finland), Svartberget (Sweden), Puéchabon (France) and Robson Creek (Australia). Test 1 yielded the most stable and informative results, with mean BT and AT differences close to 0 K and low standard deviations at most sites. The low mean differences and standard deviations indicate the correct performance of the deployed ground radiometers. For the mid-latitude stations (KIT, Puéchabon and Robson Creek) Test 2 also showed stable results with relatively constant differences throughout the year. In contrast, at the Hyytiälä and Svartberget sites, which are both located near the Arctic Circle, the low solar zenith angles (especially during winter) meant that Test 2 yielded considerable variations throughout the year. While the developed methodology needs to be further investigated over different land covers and in more arid regions, where larger differences between BT and AT may exist, it is a promising and cost-effective way to monitor the correct performance of field TIR radiometers deployed at remote sites.

Lake Surface Water Temperature (LSWT) is often considered as the reference essential climate variable for climate changes. Satellite thermal imagery has been one of the key sources of LSWT monitoring. However, accurate LSWT satellite retrieval remains challenging. In particular future high-resolution thermal Earth Observation (EO) missions, such as TRISHNA with a large viewing zenith angle and a high revisit, requires adequate in situ measurements, as well as algorithm calibration and validation. The ultimate goal of this research, conducted under the Swiss TRISHNA – Science and Electronics Contribution (T-SEC) project funded by ESA Prodex, is to improve the thermal products of upcoming TRISHNA mission and similar EO sensors for inland and coastal waters. In this study, we specifically aim at (i) assessing the effect of morphological and meteorological features on LSWT retrievals, and (ii) investigating and improving existing LSWT algorithms (e.g., Acolite-TACT, USGS-L2) based on those features. Here, we report on our existing and planned study sites in the Swiss Alps, and present the instrumentation and preliminary results for three pre- and high-alpine lakes: (1) Lake Geneva (deep large lake; 372 m a.s.l.), (2) Ägerisee (mid-size lake; 724 m a.s.l.), and (3) Steinsee (small glacier lake; 2160 m a.s.l.). Our preliminary matchup analysis between in situ measurements and Landsat 7/8/9 LSWT products looks promising. The results indicate a Mean Absolute Error (MAE) of < 1.5 °C, and a correlation coefficient of > 0.95. On the regional scale, our research will complement and profit from the ongoing lake monitoring and modeling activities in Switzerland, such as Datalakes (www.datalakes-eawag.ch), Meteolakes (http://www.meteolakes.ch), and Simstrat (www.simstrat.eawag.ch).

The TRISHNA mission (Thermal infraRed Imaging Satellite for High-resolution Natural resource Assessment) is a cooperation between the French (CNES) and Indian (ISRO) space agencies. It will measure the optical and thermal spectra emitted and reflected by the Earth from a low-altitude Sun synchronous orbit, over a swath with a width of 1026 km, approximately twice a week, at 57 m resolution for the continents and the coastal ocean. The targeted launch date for TRISHNA satellite is 2025, being then positioned as a precursor of the LSTM Copernicus mission from ESA. TRISHNA is designed for a lifetime of 5 years. Providing high-quality imagery in coastal ocean and inland waters is one of the the design drivers of the mission. Sea Surface Temperature (SST) and Lake Water Surface Temperature are Essential Climate Variables. At present, about 40% of the world’s population live within 100 km of the coast. In many regions, populations are exposed to a variety of natural hazards, as well as to the effects of global climate change, and to the impacts of human activities. Coastal zones are subject to local and remote forcings implying a wide range of phenomena, including fronts, eddies, horizontal currents, vertical velocities, plumes, tides, waves, turbulence and mixing, stratification, ice formation. Coastal marine ecosystems, such as large upwelling ecosystems, are rich and diverse, supporting much of the commercial fisheries of the world. Regarding inland waters, thermal information at fine scale is of added-value to stress the turbidity and waterborne particles. In the same regard, fine scale observations allow to assess water quality in its link with temperature, thereby bringing new insights for the productivity of biological communities, the estuary ecosystems, the halieutic resources, the detection of algal blooms and eutrophication conditions, the characterizations of marine habitats, the industrial discharge of pollutants from the rivers and estuaries into the coastal area. Improved understanding and monitoring of coastal or inland waters processes is therefore of high importance, and high resolution SST resolving fine scales of the order of 100m in coastal zones and inland waters, as expected with TRISHNA, should make an increasingly important contribution. Applications, user needs and SST retrieval challenges in coastal and inland waters will be presented.

The TRISHNA mission (Thermal infraRed Imaging Satellite for High-resolution Natural resource Assessment) is a cooperation between the French (CNES) and Indian (ISRO) space agencies. It will measure the optical and thermal spectra emitted and reflected by the Earth from a low-altitude Sun synchronous orbit, over a swath with a width of 1026 km, approximately twice a week, at 57 m resolution for the continents and the coastal ocean. The targeted launch date for TRISHNA satellite is 2025, being then positioned as a precursor of the LSTM Copernicus mission from ESA. TRISHNA is designed for a lifetime of 5 years. Providing high-quality imagery in coastal ocean and inland waters is one of the the design drivers of the mission. Retrieving and improving Sea Surface Temperature with such resolution in coastal zones is challenging, including : high variability in atmospheric water vapor, temperature and aerosol; complex shoreline, numerous islands, tides, offshore constructions; possible emissivity modification due to contaminants or high turbid waters; and turbidity in interaction with cloud detection; availability of high-quality in-situ data for optimization of retrieval algorithms and validation. In order to calibrate the SST retrieval algorithms and to validate the satellite-derived SST, satellite and in-situ reference measurements over the French coasts have been collected and databased. The assessment of the validity of the different networks (COAST-HF, ECOSCOPA, TmedNET, ISAR network, etc) for the qualification of the future TRISHNA SST coastal products will be presented.

Worldwide, the main alpine cryospheric components, such as snow, glaciers and permafrost, are undergoing drastic changes due to global climate change. The alpine cryosphere is particularly vulnerable and affected. The surface energy budget is out of balance and requires an improved monitoring, for rugged terrain in particular at the spatial length scale. It is crucial to be able to capture the individual heat fluxes and understand their spatial variability and interactions at the complex surface-atmosphere interface. Particularly key is an improved representation of all energy and mass fluxes that determine the ground thermal regime for high mountain permafrost in the first place. However, spatial monitoring of surface energy fluxes is challenging and requires imaging systems. These are characterised by various challenges to derive accurate land surface temperatures related to topography, directionality, spatial resolution and sensor specifications. Here, we present multi-sensor thermal infrared (TIR) data to obtain spatially distributed land surface temperature (LST) information of the Murtèl rockglacier in the Engadin across scales. Alongside two years of data from a terrestrial TIR camera, we obtained drone data, airborne data and point-scale radiometer and radation observations. We put a specific focus on the importance of individual processing steps for validation and calibration to obtain accurate LST data at individual scales. Our study works towards an enhanced application of thermal infrared remote sensing techniques in rugged and complex terrain, but also fosters an advancement in energy budget assessments of cryospheric components at varying spatial length scales.

Large Eurasian lakes are an integrator of climate processes at the regional scale and a good indicator of existing or potential climate changes. Variability of ice and snow regime and water dynamics is important for their physical, chemical and biological properties, and for human activity (navigation, transport, fisheries, tourism etc). We present results of our field work and satellite monitoring for ice cover and eddies under ice in lakes Baikal and Hovsgol. Multi-mission satellite observations makes it possible to monitor water dynamics and ice cover with high spatial and temporal resolution. We have used satellite imagery in the visible, near-, shortwave and thermal infrared (MODIS Terra/Aqua, Sentinel-2, Landsat 5-9 PlanetScope), active microwave observations (Sentinel-1 SAR, Jason-3 radar altimeter), historical meteorological data and data from our own dedicated field surveys and moorings. We provide qualitative and quantitative assessment of the development of currents and eddies and their horizontal and vertical structure and identify the main drivers of eddies generation. We also present how temporal analysis of ice metamorphism and evolution helps to understand and interpret the interplay influence of eddies and currents below the ice. Better understanding of eddy dynamics and continued monitoring helps to ensure safety for people travelling or working on the ice. This research was supported by the CNES TOSCA LAKEDDIES and TRISHNA, ESA CCI+ Lakes, CNRS-Russia IRN TTS and P.P. Shirshov Institute of Oceanology RAS Project N FMWE-2021-0002.

Land surface temperature (LST), latent and sensible heat fluxes are strong indicators of warming climate trends. They are affected by rising greenhouse gases (GHGs) and influence Earth’s weather and climate patterns. This is predominantly through the reduction of energy exiting Earth’s atmosphere, resulting in an increased energy budget. Key objectives for the UN Framework Convention on Climate Change (UNFCCC) investigate how Earth observations from Space could support the UNFCCC and the Paris Agreement in closing Earth’s energy budget imbalance. Improving global LST observations from satellite data to improve climate warming predictions is crucial to fulfilling this. We present the first regional trend analysis for thermal infrared LSTs with uncertainties, using a stable LST climate data record suitable for climate trend analyses. Nine representative regions including; the Amazon, Western USA, Greenland, Western Europe, The Sahel, Siberia, China, India, and Australia, were analysed using the Aqua MODIS ESA LST_cci (MYDCCI) dataset for stable climate analysis. This study highlights the importance of LST and satellite observations for monitoring surface temperature trend variability, the Earth's energy budget and its response to global warming. Through a PhD project within the National Centre of Earth Observation (NCEO) and interfacing with the ESA Climate Change Initiative Land Surface Temperature project, we aim to understand the diurnal variability in global LST better. This will be further achieved by creating the first fully integrated all-weather LST dataset that can be utilised against climate models and other temperature datasets. Here I will show some first results of understanding the merging of these LST data.

The French Mediterranean area is characterized by its high heterogeneity of land cover and topography and its frequent summer heatwaves. To mitigate drought effects on crop production and to predict forest fire danger, it is of major importance to assess the water stress of Mediterranean ecosystems, at a fine temporal and spatial scale. Future high spatial and temporal resolution thermal remote sensing missions – TRISHNA (CNES/ISRO), SBG-Thermal (NASA/ASI) and LSTM (ESA/EC) – will provide valuable data to reach these goals. Spaceborne thermal data can used to estimate the surface water stress by means of evapotranspiration (ET) models. Among thermal-based methods, the contextual ET models that rely on spatial correlations between land surface temperature (LST) and vegetation index data, have strong potential for operational applications. However, very few studies have tested such remote sensing methods over Mediterranean forests. One difficulty is related to the impact of tree cast shadows on the remotely sensed LST, which potentially hides the water stress signature. To fill the gap, this study develops a correction method to normalize the shadow effects over forests on LST-based hydric stress. We implement the Water Deficit Index (WDI) method using Landsat-7 and Landsat-8 data over a 21 km² area partially covered by a holm oak forest in South-eastern France (Puechabon). We investigate the impact of the solar zenith angle (theta) as a proxy of tree cast shadows on the satellite-retrieved WDI. In practice, the shadow effect is modelled as a linear relationship between WDI and theta depending on two parameters. The study period extends from May to September for 7 successive years (2015 to 2021) and the results are evaluated using the evaporative fraction measured in situ at the Puechabon site. The corrected WDI is more accurate than the non-corrected WDI, with a correlation coefficient (R) and root mean square error (RMSE) increasing from R=0.23 and RMSE=0.17 (no correction) to R=0.50 and RMSE= 0.12 (correction). Moreover, a method is proposed to calibrate the parameters of the correction approach on a pixel-by-pixel basis using the remotely sensed data only. We directly evaluate the linear upper hull of the WDI/theta space during particularly dry dates. The correction still improves the accuracy of WDI from a correlation coefficient of R=0.23 and RMSE=0.17 (no correction) to R=0.52 and RMSE=0.12 (correction). In the context of the near-future TRISHNA mission, this simple and self-calibrated correction brings new information to help current and future forestry challenges.

The characterization and understanding of the local and regional water cycle is primordial in the context of climate change. High temporal resolution of space-based thermal infrared (TIR) images from for example METEOSAT and MODIS, along with the development of field TIR cameras have permitted the increasing use of thermal remote sensing in Earth Sciences. TIR images are influenced by many factors such as atmosphere, solar radiation, topography and physico-chemical properties of the surface. Considering these limitations, we present several examples showing the added value of the TIR methodology to understand the subsurface hydrology dynamics at multiple spatial and temporal scales with the systematic combination of TIR images with various remote sensing data, geophysical observations and thermal/geometrical numerical modeling.. Our presentation highlights the role of subsurface fluid flows, that are controlled by permeability changes, on the surface temperature dynamic. This dynamic that ranges from meters to few hundred kilometers scale has been observed: - in civil engineering (Haropa Port quays, Normandy, France) using drone-based TIR observations; - in volcanology, within the inactive Formica Leo scoria cone and the Piton de la Fournaise volcano (La Réunion Island) using field and airborne TIR images; - in water resources, within the sedimentary Lake Chad Basin associated with surface temperature anomalies, observed from space-based images. Our studies shows that 1) ~5-10°C thermal anomalies associated to subsurface flows may be distinguished from thermal inertia/albedo/emissivity influences by taking into account the dynamics of the surface temperatures, 2) such weak thermal anomalies are observable at small scale as well at very large scale and 3) the combination of various observations and numerical modeling is very efficient to understand subsurface hydrology processes. Eventually, for the first time, high resolution spatial and temporal TIR data provided by drones to satellites bring new insights for the characterization of soil-atmosphere interactions.

The viability of agricultural production is largely dependent on the efficient use of water resources. With evapotranspiration (ET) accounting for nearly all the water used from croplands and wooded areas, accurate ET estimation methods are needed for a better understanding of irrigation demands. While surface temperature can help to detect water deficiencies, its remote sensing observation is usually influenced by the so-called directional effects, which can lead to an incorrect interpretation of observed surface emission signals. In this work, we analyse thermal radiation directionality with a modelling approach for a vineyard located in Verdu (Catalonia, Spain), using data collected in the context of the HiLiaise project. The non-continuous row site is oriented E-W. Instrumentation at the site included: net radiometers, an eddy covariance system, and thermal cameras that provided elemental soil/vegetation temperatures. To derive the overall directional surface temperatures, the measurements were aggregated by weighting the elemental values with their respective cover fractions in the viewing direction (derived using the turbid Unified Francois model or DART). The aggregated temperatures from the turbid model were compared to those from DART where correspondence was demonstrated. The reconstructed surface temperatures were then used in surface energy balance modelling schemes. Here, the soil plant atmosphere remote sensing of evapotranspiration (SPARSE) dual source model together with an extended version which discriminates shaded/unshaded elements (SPARSE4), were used to estimate the exchanges. Both schemes were able to retrieve overall fluxes satisfactorily, confirming a previous study. The sensitivity of flux and component temperature estimates to the viewing direction of the sensor was tested by using reconstructed sets of thermal data (nadir and oblique) to force the models, where we observed degradation in flux retrieval cross-row with better consistency along rows. Overall, it is nevertheless shown that by using the extended method, the sensitivity to viewing direction can significantly be reduced further off-nadir. Additionally, evaluation of output from the two-source energy balance (pyTSEB) –applied as part of the SenET programme over the broader Lleida region– show that the evapotranspiration products follow the general trend of in-situ observations. This can be explained by the relatively good agreement between the reanalysis- and the field data. Conversely, driving SPARSE/SPARSE4 with the reanalysis and other SenET input data also yields similar results to the products. To exploit strengths inherent in a variety of methods, the use of an ensemble of models in the dissemination of ET products should thus be considered.

Most hydrological, agronomical or ecological applications of any evapotranspiration or a stress factor product require an estimate for every day. However, with the projected TRISHNA revisit frequency of 3-5 days combined with the cloud interference, one needs to interpolate between two cloud free acquisitions in order to build a continuous daily evapotranspiration and water stress products. To do so, several methods have been proposed in the literature, from the simplest ones (based on easily available meteorological data) to the most complex ones (based on data assimilation of multiple sensor into distributed hydrological models). We present here the various options for an operational algorithm: (i)- an increasing complexity in accounting for the water status of the surface through water budget information (from a simple Antecedent Precipitation Index to the SAMIR model), (ii)- the various alternative sensors that can be used for cloud free days (e.g. disaggregated daily products from Sentinel 3 or MODIS such as SEN_ET) or cloudy conditions (e.g. Sentinel 1 data).

Arthropod-borne viral infections are becoming more common and may result in fatal febrile and neurological disease.  Additionally, they see no boundaries, and recently, the first domestically acquired case of dengue was recorded in the continental United States. Transmission of these viruses is seasonal and profoundly sensitive to the climate and ecological conditions driving mosquito populations and human exposure. Additionally, throughout the world, lower socio-economic status has been shown to be correlative with increased exposure to mosquitoes. Mosquito-borne infections are commonly underreported and public health interventions are reactive; thus, it is necessary to understand when and where communities are potentially at risk to proactively implement control measures. The combination of new high resolution remote sensing products along with mosquito monitoring provides the fine-scale, real-time information needed to improve our understanding of the biological process to proactively implement effective and highly targeted mosquito abatement efforts. Here, we report on our development of a spatially refined model that uses data from the ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) to capture the variability in micro-climates across the Coachella Valley, CA and incorporates them into a spatial model describing local mosquito population dynamics and viruses of concern (i.e., West Nile virus, St. Louis encephalitis and dengue). Our exposure profiles will characterize ecotonal fluctuations in mosquito habitats to identify the roles land use and climate play within mosquito development in the urban environment that are applicable for viral amplification of endemic and emerging viruses in the region and the risk zoonotic spillover to humans. Furthermore, we will characterize these ecotonal conditions in the context of the built urban environment, thermal gradients related to population dynamics, and viral amplification along with potential exposure risk related to occupational and socioeconomic status- all of which affect the risk of human zoonotic events in Coachella Valley.

Arthropod-borne viral infections are becoming more common and may result in fatal febrile and neurological disease.  Additionally, they see no boundaries, and recently, the first domestically acquired case of dengue was recorded in the continental United States. Transmission of these viruses is seasonal and profoundly sensitive to the climate and ecological conditions driving mosquito populations and human exposure. Additionally, throughout the world, lower socio-economic status has been shown to be correlative with increased exposure to mosquitoes. Mosquito-borne infections are commonly underreported and public health interventions are reactive; thus, it is necessary to understand when and where communities are potentially at risk to proactively implement control measures. The combination of new high resolution remote sensing products along with mosquito monitoring provides the fine-scale, real-time information needed to improve our understanding of the biological process to proactively implement effective and highly targeted mosquito abatement efforts. Here, we report on our development of a spatially refined model that uses data from the ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) to capture the variability in micro-climates across the Coachella Valley, CA and incorporates them into a spatial model describing local mosquito population dynamics and viruses of concern (i.e., West Nile virus, St. Louis encephalitis and dengue). Our exposure profiles will characterize ecotonal fluctuations in mosquito habitats to identify the roles land use and climate play within mosquito development in the urban environment that are applicable for viral amplification of endemic and emerging viruses in the region and the risk zoonotic spillover to humans. Furthermore, we will characterize these ecotonal conditions in the context of the built urban environment, thermal gradients related to population dynamics, and viral amplification along with potential exposure risk related to occupational and socioeconomic status- all of which affect the risk of human zoonotic events in Coachella Valley.

The TRISHNA mission (Thermal infraRed Imaging Satellite for High-resolution Natural resource Assessment), to be launched in 2025, will provide thermal infrared data with high revisit (3 acquisitions every 8 days at equator) and high spatial resolution (60 m). Such data will make it possible unprecedent monitoring of evapotranspiration and water stress. Evapotranspiration and water stress products will be proposed at level 2 within one day or less after image acquisition. We present here the various options for the operational algorithms that will be used for generating evapotranspiration and water stress products. For evapotranspiration, two main models will be used : 1- EVASPA (EVApotranspiration monitoring from SPAce, Gallego et al. 2013, Allies et al. 2020) which provides evapotranspiration maps by combining several models based on the evaporative fraction formulation of surface energy balance (contextual models) within an ensemble framework. An estimation of uncertainty in the derivation of evapotranspiration is provided by analysing the variability of the multi-model – multi-data simulations (Allies et al. 2020, Mira et al. 2016, Olioso et al. 2018) 2-STIC (Surface Temperature Initiated Closure, Mallick et al. 2014, Hu et al. 2022) which is based on the integration of radiometric temperature into a combined Penman-Monteith Shuttleworth-Wallace equation for estimating critical aerodynamic variables. The model was recently implemented within the European ECOSTRESS Hub (Hu et al. 2022). For water stress indicators, two main indices are foreseen: the evaporative fraction as provided by EVASPA and the ratio of daily evapotranspiration to reference evapotranspiration. Different methods for temporal integration of instantaneous retrievals of latent heat flux (W/m2) to daily evapotranspiration (mm/d) were proposed, a simple method scaling evapotranspiration on the basis of the instantaneous / daily solar radiation ratio showing good performances. Inputs for both models will considers other TRISHNA products (albedo, fraction cover, surface temperature and emissivity, instantaneous incoming solar radiation and atmospheric radiation) and meteorological data from meteorological analysis (air temperature, dew point temperature, daily solar and atmospheric radiations). References : Allies A., J. Demarty, et al., “Evapotranspiration Estimation in the Sahel Using a New Ensemble-Contextual Method,” Remote Sensing, 12, pp. 380, 2020. (doi:10.3390/rs12030380) Gallego-Elvira B., Olioso A., et al., “EVASPA (EVApotranspiration Assessment from SPAce) tool: An overview,” Procedia Environmental Sciences, 19, pp. 303–310, 2013 (doi: 10.1016/j.proenv.2013.06.035) Hu T., Mallick M., et al., “Evaluation of ECOSTRESS Evapotranspiration Products Retrieved from Three Structurally Contrasting Models over Europe,“ Preprint, 2022 (doi : 10.1002/essoar.10512884.1) Mallick, K., Jarvis, A.J., et al., “A Surface Temperature Initiated Closure (STIC) for surface energy balance fluxes,“ Remote Sensing of Environment, 141, pp. 243-261, 2014. Mira M., Olioso A., et al., “Uncertainty assessment of surface net radiation derived from Landsat images,” Remote sensing of Environment, 175, pp. 251–270, 2016 (doi: 10.1016/j.rse.2015.12.054) Olioso A., Allies A., et al., “Monitoring evapotranspiration from remote sensing data and ground data using ensemble model averaging,” IGARSS2018, 23-27 juillet 2018, Valencia, España, pp. 7656-7659, 2018 (doi: 10.1109/IGARSS.2018.8517532)

Thermal remote sensing has emerged as a powerful tool for capturing spatiotemporal dynamics of ecosystem processes at different scales. In this study, we present two long-term in-situ thermal datasets: a mixed temperate forest in Massachusetts (Harvard Forest) and subalpine conifer forest in Colorado (Niwot Ridge). We validated the accuracy of camera-derived temperatures against thermocouples, but identified calibration drift over time that requires accounting for. Accurate temperature measurements require consideration of emissivity variations of plant leaves due to factors such as leaf age, water content, and surface roughness. Our dataset can be utilized to evaluate the accuracy and effectiveness of new remote sensing products, leading to more reliable and precise estimates of ecosystem processes at larger scales. Specifically, by comparing our measured temperature data with model outputs or satellite-based temperature estimates, we can identify and address discrepancies, improve our understanding of ecosystem processes and their response to environmental drivers.

Thermal signals can be detected across a wide range of the electromagnetic spectrum. The wavebands selected for thermal earth observation missions have various trade-offs (range of detectable temperature, resolution, sensitivity), but also exhibit various properties (emissivity constraints). As a result, interoperability between wavebands in thermal imagery is complex, as these properties affect imagery in different ways. The MODIS/ASTER Airborne Simulator (MASTER) has been operational since 1998 and collects data across a range of 50 wavebands covering thermal bands on the MODIS and ASTER satellites. Satellite Vu have been using MASTER to explore relationships between these thermal bands with the goal of enhancing interoperability between them in advance of our first medium wave infrared (MWIR) sensor due to launch in 2023. Within this presentation, we will introduce the MASTER mission, it’s spatio-temporal coverage, and experiments which Satellite Vu have been running using MASTER data to compare the MWIR and long wave infrared (present on ASTER, Landsat and ECOSTRESS satellites) channels. We will discuss some of the complexities of imaging within the MWIR, which we can see within the MASTER datasets (for example, atmospheric interference during daytime imaging), and how Satellite Vu have been preparing to handle these challenges once in orbit.

The SBG-TIR Project contains two instruments (TIR and VNIR) that work in concert to achieve and expand upon the scientific objectives laid out in the Decadal Survey. The TIR instrument (OTTER: Orbiting Terrestrial Thermal Emission Radiometer), delivered by JPL, is uniquely situated to continue the legacy initiated by ECOSTRESS while taking advantage of recent technological advances to expand the coverage and capabilities. The instrument and mission design will enable global coverage every 3-days. This is enhanced by the features of additional observatories from different international partners. This talk will explore the progress and design of the OTTER Instrument and describe the trades performed in concert with our ASI SBG-TIR partners.

Applications involving observation of the Earth’s surface from satellite platforms on a lower than regional scale, such as crop monitoring, require greater availability of thermal information, in particular land surface temperature (LST), with spatial resolutions appropriate for local studies. Therefore, numerous authors have proposed and developed methods to extract LST at the “subpixel” level, through the use of complementary remote sensing products, with results suitable at higher resolutions. Most of these methods are based on the correlation between vegetation indices, as is the case of the Normalized Difference Vegetation Index (NDVI), and LST, for land covers with specific characteristics. These methods are based on the implementation of “traditional” statistical models, such as linear or quadratic regressions. The availability of other vegetation indices or indices related to water availability has enormous potential, thanks to the contribution offered by possible new estimators. This fact, together with the development of advanced computing methods, based on machine learning techniques, can lead to create more robust disaggregation algorithms. This study analyzes the behavior and contribution of several spectral indices, as well as other complementary variables, for the development of advanced models, which have their origin in the field of Artificial Intelligence. Through these models, it is intended to bring the original resolution (regional scale) of the dependent variable LST to the local scale. In particular, we generate LST maps at the high resolution of the MSI sensor (20 m), on board the Sentinel 2 platforms, starting from the moderate resolution of the thermal bands of EOS-MODIS sensor (1000 m). This contribution shows the first results obtained by applying these disaggregation methodologies with different variables, both spectral and of other nature. The study had the financial support of the project Tool4Extreme PID2020-118797RBI00 funded by MCIN/AEI/10.13039/501100011033.

Introduction Very few minutes can decide how fast fire fighters can gain control over a forest fire, and which damage to humans, nature, and economy it causes. Thermal data of satellites is thereby extremely valuable, as fires can be detected and monitored in large areas independent of factors like wind-speed, terrain or day-light. However, the fire information often reaches the fire fighters with a delay of up to one hour, as the data must be first downlinked at the next ground station and then processed on-ground. To mitigate that, the Munich-based company OroraTech presents the concept of on-orbit fire detection within a CubeSat constellation. We show first results and learnings on our satellite FOREST-1. Concept To reduce the time between imaging and informing on-ground personnel, the fire detection is performed directly on the satellite. The fire coordinates are compressed in a tiny file of a few kilobytes, which is then sent via satellite-to-satellite communication to the ground and forwarded to the fire department control center. This reduces the time between image acquisition and action to a few minutes. Success on Forest-1 OroraTech launched its first satellite FOREST-1 in early 2022 for a technological demonstration of on-orbit wildfire detection. Wildfires around the globe were successfully imaged and used to test and adjust state-of-the-art fire detection methods. One well-performing method was chosen, and uplinked to Forest-1. We showed that selected fires were successfully detected on-orbit. After those first successes with FOREST-1, further development up to the end-to-end test from data recording to on-site personnel will happen within 2023 with the second satellite FOREST-2.

During the last two decades, much progress has been done in volcano remote sensing, and further developments are expected over the next few years considering the new planned space missions: ESA Sentinel satellites, TRISHNA (CNES/ISRO), SBG-Thermal (NASA/ASI), LSTM (ESA/EC) and other mission studies that include thermal infrared sensors. Specifically, the channel configuration of SBG-Thermal has been investigated for volcanic applications with particular attention to high temperature events and volcanic gas emissions. In fact, the five channels in the thermal spectral region and the two ones in the middle-wave infrared allow estimating the surface temperature in the range 300-1200 K. In this work, we perform a theoretical study on the capability of the SBG-Thermal instrument to detect subtle thermal anomalies related to volcanic activity and investigate its potentiality to be used for effusion rate estimation in case of effusive eruptions. Furthermore, we explore the chance to employ the channel at 4.8 μm analyzed for CO2 estimation purposes. We finally demonstrate our theoretical approach by using the MIR-TIR data (channels 31 peaked at 3.9 μm in conjunction with channels 47 and 48 peaked at 10.63 μm and 11.32 μm and the channel 37 peaked at 4.8 μm) acquired by the MASTER (Modis/ASTER) instrument during the 2018 Kilauea eruption.

During explosive eruptions, volcanoes can inject into the atmosphere ash, water vapor and different gasses (like SO2, CO2, etc..), which produce volcanic clouds. They can spread over a great distance and remain in the atmosphere for a very long time. Monitoring volcanic ash clouds is beneficial to implement risk mitigation measures and preventing volcanic crisis from becoming disasters. With the rapid development of Earth observation technology, a variety of satellite data in different spectral ranges with diverse spatial and temporal resolutions are well suitable to monitor in global scale volcanic clouds in an efficient and timely manner. For this reason, the integration of different satellite data makes possible a continuous monitoring of a volcanic explosive eruption. Here, we analyzed multispectral images using artificial intelligence (AI) techniques, in order to track the evolution of a volcanic cloud and therefore to understand which regions may be most affected by its impact. Specifically, we have developed an algorithm with the objective of: (i) identifying and isolating a volcanic cloud; (ii) characterizing its main components; (iii) determining the directions spread of the volcanic cloud. The techniques employed to implement this algorithm are based on machine learning (ML), such as support vector machine (SVM) and random forest (RF), and image processing approaches, such as Thermal Image Velocimetry (TIV). This AI model was applied to different satellite instruments, in order to perform a near real-time monitoring of the volcanic clouds emitted during some of the paroxysmal explosive events that occurred at Mt. Etna between 2020 and 2022.

Volcano hazard monitoring aims to determine where and when future volcano hazards will occur and their potential severity. By monitoring, we mean both following the manifestations of the eruption once it has started, as well as forecasting the areas potentially threatened by hazardous phenomena, producing different scenarios as eruptive conditions change. Here, we propose an emerging strategy for volcanic hazard monitoring based on the integration of satellite remote sensing techniques and innovative Artificial Intelligence (AI) models for detecting, measuring and tracking eruptive phenomena. Satellite remote sensing can yield an improved understanding of volcanic processes and volcanic hazards simply by providing more frequent observations at a wide variety of wavelengths. The increasing availability of open-source satellite data and current developments in cloud computing and data-driven approaches have made the monitoring of volcanic hazards from space more feasible for volcano observatories. We developed an AI-based platform to monitor in near real-time different volcanic hazardous phenomena using thermal satellite images. Several built-in modules cope together towards a common goal, i.e., detecting the onset of the eruption and following the manifestations of the volcanic activity once it starts. Advanced ML algorithms are used to retrieve information about the ongoing volcanic activity in time and space. Under this perspective, machine learning (ML), a type of AI in which computers learn from data, is gaining importance in volcanology, not only for monitoring purposes (i.e., in real-time) but also for subsequent hazards analysis (e.g. modeling tools). The collection of models and methods includes advanced satellite techniques for ash plumes and lava flows identification and characterization, coupled with AI models for real-time scenario forecasting and volcanic hazard assessment. We will describe and demonstrate the operation of this AI-based platform during some recent eruptive events at Stromboli volcano (Sicily, Italy).

The aim of this work was to study what thermal infrared time series can bring to anomaly detection or land classification. During the study, we created thermal infrared pixel time series from Landsat 8 band 10 on two main use cases. We used the « Landsat 8-9 Collection 2 Level 2 Science » dataset from the USGS platform. The pixel resolution is 100m, the revisit is 16 days. The first use case was the Yinchuan region in China, which is frequently affected by floods. The second use case was the La Palma volcano in the Canary Islands which suffered volcanic eruptions in September 2021. The first study consisted of using Sarima (seaonal autoregressive integrated moving average) algorithms on the pixel time series to precisely detect (at pixel level) the lava flow zones for the La Palma volcano. We used the SARIMA algorithms to model the normal behavior of the surface temperature of the pixel around the volcano. Then, we detected anomalies when the actual surface temperature calculated from the Landsat 8 data was different from the Sarima prediction. The second study consisted of using clustering algorithms (Time Series K Means) on the pixel surface temperature time series to detect and classify flood zones on the Yinchuan use case. We used the Time Series KMeans algorithms to create a mask of two classes on the use case: the water class which corresponds to the pixels in water state most of the time and the land class which corresponds to the pixels in land state. For each date of the time serie we calculated the mean of each class and we compared each pixel to these means to classify them as “water” or “land”. If the percentage of the “water” pixels was higher than the normal, we considered the date as “flooded”.

Fire radiative power (FRP) is well related to rates of fuel consumption and smoke emission. The FRP of active fires (AFs) is routinely assessed with spaceborne sensors such as Meteosat SEVIRI, MODIS, VIIRS and SLSTR, and used in many scientific and operational applications worldwide. However, spaceborne sensors do not currently detect all potentially detectable active fires, even if they are burning under cloud free conditions, which leads to underestimates in the amount of fire that is estimated to be occurring and in the amount of carbon, particulates and trace gases calculated to be released. MODIS for example has a 1 km nadir pixel size that provides a minimum per-pixel FRP detection limit of ~5–8 MW, leading to significant undercounting of AF pixels with FRPs of less than around 10 MW. SLSTR by night offers somewhat better performance than this, whereas by day it is probably somewhat worse. Low FRP AF pixels are in fact the most common type, and undercounting with geostationary sensors is even more significant than with polar orbiters - though they do provide the advantage of almost continuous observations of rapidly changing fire situations. Conversely however, missing low FRP AF pixels may not be too significant for overall total FRP determination, since each missed pixel contains only a very limited amount of fire. The exact magnitude of the landscape-scale FRP underestimation induced by AF undercounting still remains poorly understood overall, as does how it varies with sensor pixel size and overpass time. This presentation will show evidence of the phenomena, will use airborne and other data to investigate these issues, and will contain recommendations that can help guide future satellite sensor design where active fire detection and FRP retrieval is targeted. TRANSLATE with x English ArabicHebrewPolish BulgarianHindiPortuguese CatalanHmong DawRomanian Chinese SimplifiedHungarianRussian Chinese TraditionalIndonesianSlovak CzechItalianSlovenian DanishJapaneseSpanish DutchKlingonSwedish EnglishKoreanThai EstonianLatvianTurkish FinnishLithuanianUkrainian FrenchMalayUrdu GermanMalteseVietnamese GreekNorwegianWelsh Haitian CreolePersian TRANSLATE with COPY THE URL BELOW Back EMBED THE SNIPPET BELOW IN YOUR SITE Enable collaborative features and customize widget: Bing Webmaster Portal Back TRANSLATE with x English ArabicHebrewPolish BulgarianHindiPortuguese CatalanHmong DawRomanian Chinese SimplifiedHungarianRussian Chinese TraditionalIndonesianSlovak CzechItalianSlovenian DanishJapaneseSpanish DutchKlingonSwedish EnglishKoreanThai EstonianLatvianTurkish FinnishLithuanianUkrainian FrenchMalayUrdu GermanMalteseVietnamese GreekNorwegianWelsh Haitian CreolePersian TRANSLATE with COPY THE URL BELOW Back EMBED THE SNIPPET BELOW IN YOUR SITE Enable collaborative features and customize widget: Bing Webmaster Portal Back

NASA’s thermal ECOSTRESS mission was originally designed to measure evaporative plant stress on a near-global scale. In the GeoHot project we are using these data for geologic applications, namely to investigate high-enthalpy geothermal resources, a vital source in the global energy transition. The aim of this project is to optimize the geothermal temperature anomaly detection from space by using a different and innovative approaches. GeoHot is supported by the Dutch Research Council’s User Support Programme Space Research (NWO-GO) as well as the NASA ECOSTRESS Science and Application Team Grant, and runs from 2021 to 2024. The first two years of the project have focused on assessing the suitability of the ECOSTRESS data for the intended application, which requires time series of nighttime data with high geometric and radiometric fidelity. To achieve that, an novel matching approach between nighttime ECOSTRESS TIR images and a SENTINEL land cover classification of water bodies was developed. Secondly, we creatwed a non-standard data processing chain that reduces chess-board like textures that are caused by radiometric differences in the overlapping parts of the scan lines of the rotating mirror. We then developed an algorithm for detecting anomalously warm geothermal pixels as compared to the pixels immediate surroundings. We tested the anomaly detection algorithm on the geothermal area of Olkaria, Kenya, and conducted fieldwork to validate the detections results against known fumaroles and newly detected areas with elevated surface temperatures. (overall accuracy around 78%). In this talk we will look back at lessons learnt during the first 2 years of GeoHot, as well as forward to the plans for the year ahead, as well as for future TIR missions that may have implications for geothermal energy exploration.

Climate change is causing increasingly severe environmental and economic damages and we will have to adapt in many fields of our daily life. City planning will have to consider the overall rising temperatures in certain parts of urban areas, and agriculture will have to reduce the amount of water consumption used for irrigating fields. Simultaneously, the occurrence of wildfires will drastically increase in parts of the world as already observed in the last few years. Satellite-based thermal infrared (TIR) data can help in mitigating and monitoring these problems, however, the temporal and spatial resolution of currently available TIR data is not sufficient for many of these use cases. At OroraTech, we aim at developing a constellation of TIR sensors with a spatial resolution of 200 m and revisit time of 30 minutes worldwide, complementing data from existing LEO and GEO missions. Here, we present images as well as time series results of various places on the globe of our first satellite mission FOREST-1, a technology demonstrator launched in January 2022. We successfully imaged several hundred target scenes on Earth in long-wave and mid-wave IR and detected dozens of active wildfires. The lessons learned of FOREST-1 significantly shaped the design of our 2nd generation TIR imager FOREST-2, which will be launched in mid 2023.

Variations in silicate/quartz mineralogy are particularly useful for geologic mapping because they are an essential criterion for tracing magmatic fluids and targeting Au ± Cu mineralization. Hyperspectral thermal infrared images can capture the diagnostic absorption features of quartz, a previously unmapped mineral when using only SWIR (Shortwave Infrared) images. This study aims to map quartz concentrations using hyperspectral thermal infrared images at the laboratory as well as airborne scale. A spectral index was developed using the depth of quartz doublet absorption features (spectral quartz index, SQI) and linked back to the concentration of quartz. Threshold values for each interval are determined from synthetic linear mixtures datasets of quartz mixed with alunite and pyrophyllite. These mixtures were chosen as these minerals occur together in the central part of an epithermally altered system, and their thermal infrared spectral features partially overlap. The approach was assessed on different scales, from rock samples in the laboratory (OWL, 400 μm pixel spacing) to airborne (SEBASS, 1 m pixel spacing) over the epithermally altered “Alunite Hill”, Yerington district, Nevada, USA. For rock samples, the SQI classified images were visually compared with QEMSCAN images, and the spectral quartz abundance from laboratory images was linearly correlated with QEMSCAN-based quartz abundance. Quartz abundance derived from airborne imagery was compared to laboratory-derived quartz abundance by averaging 3x3 airborne pixels centered on the field sample location to partially compensate for the geolocation uncertainty. Results indicate the linear correlation of SQI from laboratory and airborne, with.R2 = 0.4 and an average error of 1.37% and. The quartz-rich zones identified by this method are consistent with advanced argillic alterations in the geological maps of the study area and have the potential to remotely map zones with the emplacement of magmatic-hydrothermal fluids.

The ERC Synergy (ERC-SyG) Project urbisphere aims to forecast feedbacks between weather/climate and cities, by exploiting new synergies between spatial planning, remote sensing, modelling and ground-based observations, and incorporating city dynamics and human behaviour into weather and climate forecasts/projections. The urbisphere field campaign in Berlin, Germany, provides new information on the impact of cities on the urban- and regional-scale boundary layer using data measured across a wide range of scales during the course of a full year (Autumn 2021 to Autumn 2022). During an intensive thermal infrared (TIR) observation campaign in August 2022, sensors included three TIR cameras (Optris 640 Pi and Optris 400 Pi) mounted on the ground and a building roof, SatelliteVu MIR (Mid-Infrared) sensor mounted on an aircraft, and Anafi Parrot Thermal sensor mounted on UAV (Unmanned Aerial Vehicle). This was completed with satellite observations from Sentinel-3 SLSTR, MODIS, ASTER, ECOSTRESS and Landsat. Thus, the Intensive observation period (IOP) has a wide range of spatial resolutions (<1 m to 1 km), many collected over the same location and many at the same time. The sensors differ in the field of view, their wavelength, and their accuracy. In this contribution, we provide an overview of the TIR and MIR observations, their spatial and temporal coverages, and initial results for evaluating the spatial and temporal variability of surface temperature during the IOP. Acknowledgement This work is part of the urbisphere project (www.urbisphere.eu), a synergy project funded by the European Research Council (ERC-SyG) within the European Union’s Horizon 2020 research and innovation program under grant agreement no. 855005. Special thanks to the Chair of Climatology at Technische Universität Berlin for providing equipment, ensuring access to observation sites and to all those who contributed to the field work: Fred Meier, Kai König, Josefine Brückmann.

Satellite thermal images have been using for several years in geological research fields. During the last two decades, much progress has been done in remote sensing techniques, and further substantive developments are expected over the next few years considering the new space missions: incoming ESA Sentinel satellites, last launched NASA Landsat-9, planned NASA-SBG, CNES TRISHNA missions and other mission studies that include thermal infrared sensors. One of the main applications of thermal images regards the Land Surface Temperature (LST) and Emissivity analysis. In this work, we consider the foreseen channel configuration employed in the future mission SBG-TIR managed by NASA-ASI. Specifically, the five channels in the thermal spectral region and the two in the middle-wave infrared allow estimating the surface temperature in the range 300-1200 K. The SBG-TIR mission can be considered a very performant successor of the actual operative TIR missions such as ASTER (from 1999 on Terra satellite) and ECOSTRESS (from 2018 onboard ISS). Concerning geological applications, we envisage for the TIR multispectral data set, also in synergy with hyperspectral, two potential applications: raw material (e.g. metamorphic silica formations like serpentine) and Soil Organic Content (SOC) of agricultural topsoil. In this context, we intend to explore LST and emissivity data set, derived from ECOSTRESS data set or resampled from airborne TASI-600 surveys, on specific Italian geologic framework and relevant agricultural test sites to analyze the SBG-TIR potential within these scientific topics.

There is great interest in improving forecast meteorology for urban areas, particularly in regards to surface temperature, which often shows considerable within-city variability. This hetereogeneity presents an observational challenge for urban numerical weather prediction (UNWP), to which high (~<100 m) resolution thermal observations can contribute. We investigated the extent to which Landsat 8/9 data can usefully inform assessments of a Met Office 100-m scale UNWP model for London (UK). Comparisons for clear-sky days rapidly identified aspects of the auxiliary data used within the UNWP that correlated with model surface temperature errors, which was inferred because of these aspects' spatial correlations with model-Landsat differences. Larger scale differences between UNWP and observations are more ambiguous, as part of these may be definitional, relating to the differences in the surface temperature "seen" from space compared to the meaning of the nearest equivalent model variable.

With approximately 50% of people worldwide living in urbanised areas, it is more important than ever to consider how rising temperatures will affect our cities and the health of those living in them. An important step in doing this is to analyse the urban heat island (UHI) effect which states that an area of industrial or urban cover that suffers generally higher temperatures than neighbouring rural regions. This paper considers the UHI within the urban extent that surrounds Birmingham, UK by looking at land surface temperatures (LST) from the Landsat 8 mission, at 100m spatial resolution in the thermal bands. The LST is derived using a bespoke optimal estimation technique developed at the University of Leicester. A rural background reference was created through combination of four regions, using Normalized Difference Vegetation Index (NDVI) to determine vegetation content, these regions were analysed to determine the variability and thereby to ensure a robust and well defined rural back subtraction for the UHI. By considering the 90th percentile of pixels within the city centre region, results show that during February 2021 the city centre experiences an increase in temperature of 3.2 ± 1.8°C. In July of the same year, the increase in temperature rises up to 11.7 ± 2.5°C. The suburban region experiences an analogous, yet lessened effect with temperatures rising between 1.8 ± 1.6°C – 8.2 ± 2.4°C when compared with the rural background. This study has additionally investigated the estimation of a thermal discomfort or “feels-like” temperature.

Responding to global warming and adapting to climate change effects such as heat waves and drought is a key priority of European and national-level Climate Change Adaptation strategies. Regional and city administrations aim to reduce climate-change-related health risks and to increase human well-being by adequate planning measures such as establishing green and blue infrastructure. Changes in land use (LU) and land cover (LC) play an important role in determining local climate characteristics. Urban Climate, for instance, differs from the surrounding natural areas, showing higher air and surface temperatures, known as the Urban Heat Island Effect, mainly related to changes of the surface radiative properties. These modifications in the built-up environment make cities warmer than their surroundings and more prone to excess heat. Global and regional warming can further amplify the effect of excess heat. Understanding how land use and climate trends lead to changes to the local climate is essential for decision-makers to find optimally cost-effective, evidence-based, and consistent solutions for sustainable cities and communities. Therefore, we have started an activity that will combine LULC (e.g., Copernicus Land Monitoring Service) and climate data (from regional-scale climate modelling) as well as EO-based land surface temperature observations captured at various resolutions (e.g., MODIS/VIIRS, Sentinel-3, Landsat, ASTER, ECOSTRESS) to demonstrate the effect of urbanization or other LULC changes on ambient temperatures at high spatial resolution (<50m) applying a multi-sensor/data multi-resolution downscalling algorithm. Making use of RCM-based scenario data will further allow to assess future expected temperature increases and heat impact to identify potential hotspot areas for which adaptation actions will be required. Combining these datasets allows to develop a framework to bring RCM data down to the city-block scale in order to establish a decision tool for communal spatial planning units. Since taking action in terms of adaptation is not only the focus of larger cities but also of many smaller urban communities, such a tool will also be of particular interest for small to medium size urban centers allowing to find optimally cost-effective, evidence-based, and consistent solutions for sustainable municipalities.

Due to climate change, the intensity and frequency of heat waves is projected to increase in the near future. During these events, citizens might experience thermal stress, which negatively impacts their health, ultimately leading to an increase in mortality and morbidity rates. Although urban population is at a higher risk due to the heat island effect, the physical and mental well-being of both rural and urban communities is affected during heat waves. Wallonia, South Belgium, is characterized by a growing sub-urban population and a decrease in inhabitants in city centers. It is essential to account for this regional characteristic when developing adaptations to make the territory resilient to future changes, especially to thermal hazard. To plan actions, public authorities need spatial information about heat health risks, which is influenced by the population exposure to thermal hazard and the population vulnerability. In this context, the objective of this research is to investigate the potential of land surface temperature (LST) measured by thermal remote sensing to map the heat health risk associated with thermal hazard at high spatial resolution and regional scale in Wallonia. The project will be articulated in four phases and will be developed in close collaboration with relevant stakeholder and public citizens. First, a regional time-series of LST will be developed using available thermal satellite data such as Landsat, Aster and Sentinel 3. In the second phase, the quality and the LST time-series potential to be used as a basis to calculate the thermal heat health risk will be evaluated. Using land cover maps, the influence of different land cover types on the LST values will be analyzed at regional scales. The LST values will also be compared to meteorological data from weather stations evenly distributed in Wallonia. This analysis will inform us about the correlation between satellite-measured LST and both air temperature and thermal indices. During the third phase, the heat health risk will be estimated at regional scale. This will be based on a known methodology and will include the mapping of the thermal hazard, the population exposure and vulnerability to thermal stress and will use the LST time series as input. The fourth and final phase of this project will focus on developing decision-making tools to help policy makers in the transition towards a more resilient territory. The tools will be defined in a collaborative process through thematic workshops involving different stakeholders.

How cities have been designed, constructed and managed alters their temperature leading to Urban Heat Island (UHI) impact. Land surface temperature (LST) is a key parameter for estimating Surface urban heat island intensity (SUHII). In recent decades, UHI mapping and modelling have been one of the most active areas of research due to the accessibility and advancement of satellite remote sensing imagery. The heat experienced within microclimates and the regional UHI impact have a complicated relationship. Therefore, spatial resolution of the data plays an important role when making plans for heat mitigation at various scales, it is important to take into account both the larger UHI effect and microclimates. This paper aims to understand the effect of spatial resolution of thermal data on the estimation of SUHII. LST from MODIS and Landsat satellites were used to prepare SUHII maps over the city of Navi Mumbai in India. In this study, SUHII was defined as the difference between the LST of any pixel and the LST of urban vegetation within that city. It was observed that the SUHII value derived from Landsat had more dynamic range compared to MODIS. Also, the SUHII derived from MODIS was not able to locate particular hot spots and cool spots within the city, resulting in a misinformation on the thermal nature of different zones in the city. The presence of small scale heat or cool pockets within the city were not identified in the MODIS SUHII maps and it can be used only for a regional heat island analysis. Literature suggest that focusing on mitigation measures from the UHI effect emphases on proposing measures at a local scale followed by adapting them on a regional scale. For an efficient urban planning interventions towards city cooling, high-resolution thermal sensors are required to obtain data at 30 m to 50 m pixel size.

Land Surface Temperature (LST) is an important indicator for assessing the impacts of global warming, land use change, and human-environment interaction, as well as hydrological processes and climate change studies. Monitoring and managing ecological changes in vegetation and land can be obtained by analyzing the area's past and present Land Use and Land Cover (LULC) categories. LST was retrieved using Thermal Infrared Sensors (TIRS) by the Single-Channel (SC) algorithm for Landsat-7 Enhanced Thematic Mapper Plus (ETM+) and the Split-Window (SW) algorithm for Landsat-8 Operational Land Imager (OLI) across the watershed for the period 2001 to 2021. The spatiotemporal changes in LULC and Normalized Difference Vegetation Index (NDVI) were retrieved, along with land surface emissivity (LSE), using geo-spatial techniques. In both Landsat datasets, the supervised classification method employing the Support Vector Machine (SVM) algorithm was utilized. There was a significant decline of -3.56% in glaciers/snow, -0.60% in Himalayan moist temperate forests, and -0.47% in Himalayan dry temperate forests. On the other hand, the study revealed about a 0.07%, 0.71%, 1.21%, 1.04%, and 1.30% increase in the area of built-up/settlements, agricultural/plantations, vegetation/grasslands/grazing lands, barren/sandy, and rocky/open/debris lands, respectively, during the specified time period. In 2001, the spatial distribution of LST was between a low of -05.59°C and a high of 34.60°C, whereas, in 2021, these values varied between a low of -06.57°C and a high of 35.49°C. The relative comparison of LST on various LULC categories, derived from SC and SW algorithms, showed that there was an average difference of ± 1°C from 2001 to 2021. As a result, we hypothesize that the primary drivers of LULC that influence the LST changes in the Parbati Watershed are population growth, rapid developmental and anthropogenic activities, as well as unconstrained tourism growth. This investigation will obtain scientific information on the origins of extreme LST and potential mitigation strategies. Policymakers could make use of this research to develop capabilities that increase the hilly landscape’s long-term effectiveness.

Surface urban heat islands (SUHI) have particular relevance as temperature increase directly affects population health and comfort. Remote sensing data have been widely used in the last decades in urban climate studies, with datasets being available in various temporal and spatial resolutions. A majority of remote sensing SUHI studies rely on polar orbiting satellites. These studies have greatly improved our understanding of SUHI, especially its trends and seasonal variability. Work based on these sensors have even analyzed the diurnal cycle of SUHI, however relying on few daily observations (two to four observations, at best). As the revisit time of polar orbiting sensors may not be enough to characterize SUHIs diurnal cycle, geostationary satellites have been used to combat this limitation. Some work has already been done into downscaling land surface temperature (LST) from geostationary sensors, in order to better understand both temporal and spatial variability of SUHI. We aim to use the differences between sensor spatiotemporal characteristics as an asset in analyzing the SUHI effect in three cities (Paris, Madrid and Milan). SUHI was computed based on LST retrieved from one geostationary and two higher resolution sensors: the Spinning Enhanced Visible and Infrared Imager (SEVIRI) onboard MSG, the Advanced Very High Resolution Radiometer (AVHRR) onboard Metop and the Visible Infrared Imaging Radiometer Suite (VIIRS) onboard Suomi NPP. The study was conducted for the period of 2015 – 2022 with the aim of identifying the added value of combining high spatial with high temporal resolution data.

The risk and frequency of heatwaves is rising due to anthropogenic climate change specifically for urban cities effected by urban heat island. Nature based solutions have been used as a mitigation strategy for various urban issues including urban heat island effect. The UPSURGE project aims to use Nature based solutions for regenerative development in five demonstration cities. The five cities are based in different climate zones, vary in population, consists of single to multiple demonstration sites, and are deploying various Nature based solutions based on the key challenges. The demonstration cities include Belfast, Breda, Budapest, Maribor, and Katowice. The demonstration sites are being Co-designed with stakeholders to address local concerns, diverse perspectives and involve citizens to address the longevity of Nature based solutions. The cities have selected Nature based solutions varying from green roof, green wall, Miyawaki forest, raingardens, agroecology community gardens, rewilded zones, meadows, climate arboretum, water gardens. The work aims to analyse the surface urban heat island effect of these five demo cities during heatwave. The local climate zone approach is used to understand the neighbourhoods within the demo cities. The variation in urban heat is analysed utilizing the probability distribution of land surface temperature. The LST data from Sentinel-3 and Ecostress have been used using Google Earth Engine. The LST data over the last 5 year during heatwaves have been analysed to understand the effect of Covid lockdowns. The Kullback- Leibler divergence statistics is estimated to determine the distance between the distributions of LST data. The most vulnerable neighbourhoods in each demo city have been highlighted having the highest probability and maximum statistical distance. The surface urban heat island significantly reduced during the Covid period.

Landsat TRS Tools is an ArcGIS Desktop 10+ toolbox for automatic retrieval of brightness temperature (BT), land surface emissivity (LSE) and land surface temperature (LST) from LANDSAT data. It consists of three separate tools written in Python scripting language. The main objective to develop the toolbox was to increase the efficiency of processing Landsat satellite data to estimate LST patterns in the urban area of Krakow, the second largest city of Poland. These studies were carried out in the Satellite Remote Sensing Centre, Institute of Meteorology and Water Management – National Research Institute (IMGW-PIB) in collaboration with the private company ESRI Polska (an authorised Polish distributor of ArcGIS software). It was my own enthusiastic bottom-up initiative, as an early career researcher, to start these research activities and establish a long-term collaboration between the Satellite Remote Sensing Centre, IMGW-PIB and ESRI Polska for the benefit of both parties. In July 2012 we presented our Landsat TRS tools toolbox and its functionalities for a wider audience during the IEEE Geoscience and Remote Sensing Symposium (IGARSS) in Munich. After the conference we published a paper about the Landsat TRS tools concept (Walawender et al 2012), in which we declared our openness to share the toolbox at no cost for all kinds of scientific activities. Authorisation was based on a short questionnaire providing us with basic information on the scope of the project for which the toolbox will be used. In October 2016 I quit my job at IMGW-PIB and moved to Germany. There has been no further development of the Landsat TRS tools since then, although I continued to respond to the toolbox requests. Last year marked 10 years since the release of Landsat TRS tools. The list of authorised users reached a number of 110 people from 42 countries around the world, which used the toolbox for a wide range of different environmental applications. This presentation wraps up all these applications in the form of short statistical analysis based on the information from the user questionnaires. It turned out that our toolbox might have played an unexpected role in building capacity among the Landsat data users with less experience in satellite data processing and retrieval algorithms, often based in developing countries. Implementing automatic processing of Landsat data in the GIS environment enabled easy integration of the satellite-based products with other datasets from various different sources, followed by a quick spatial analysis to support decision-making processes. Walawender, J.P., Hajto, M.J. and Iwaniuk, P. (2012), A new ArcGIS toolset for automated mapping of land surface temperature with the use of LANDSAT satellite data, Proc. IEEE IGARSS, 22–27 July 2012, Munich, Germany, 4371–4374, doi: 10.1109/IGARSS.2012.6350405

With increasing heat waves frequency, cities will face major environmental and public health issues. In the coming years, the future thermal infrared (TIR) satellite missions will make it possible to produce high spatial and temporal resolution Land Surface Temperature (LST) maps in order to better understand the urban heat island phenomenon at district or city scale. However, to accurately retrieve LST from satellite data, it is crucial to get a good understanding of the TIR radiative interactions at canyon scale. In that respect, 3D urban thermo-radiative models are valuable tools as they can simulate at very fine scale the radiative exchanges in the urban canopy. But, prior of using them, it is necessary to validate them. To do so, we set up an experiment dedicated to the validation of 3D physical simulation using in-situ sensors and remote sensing observations. This experiment took place during the CAMCATT-AI4GEO field campaign led in Toulouse city in June 2021. Several buildings were instrumented with iButtons, completed with a thermal infrared camera. In addition, various radiometers were used to collect the optical properties of the material of the study site. A side experiment was carried out to evaluate the iButtons data using KT19 radiometers as reference. After confirmation of the reliability of the acquired iButton dataset, we compared the data collected during the experiment with LST simulated by the SOLENE-microclimat urban microclimate model. This presentation first describes the experiment set-up, the collected dataset and the SOLENE-simulations set up for the studied scene. Next, it presents the comparison between the retrieved and simulated LST. Finally, the accuracy of the model is investigated and discussed taking into account the sensors reliability.

The CAMCATT-AI4GEO extensive field experiment took place in Toulouse from 14 to 25 June 2021. Its main objective was the acquisition of a new reference dataset on an urban site to support the development and validation of data products for the future TRISHNA mission. This field campaign led to a unique set of data combining airborne VISNIR-SWIR hyperspectral imagery, multispectral thermal infrared imagery and 3D LiDAR acquisitions along with various ground data collected, for some of them, simultaneously to the flight. The ground-based dataset comprises surface reflectance measured spectrally using ASD spectroradiometers as well as in 6 spectral bands spreading from shortwave to thermal infrared and for two observation angles with a SOC410-DHR handheld reflectometer. It is completed with land surface temperature (LST) and emissivity (LSE) retrieved from thermal infrared radiance acquired in 6 spectral bands using CIMEL radiometers. It also includes meteorological data coming from 4 radiosoundings performed during the flights, data routinely collected at the Blagnac airport reference station as well as air temperature and humidity acquired using instrumented cars following two different itineraries. As initially intended, this dataset will allow the validation of at-surface radiance, LST and LSE data products as well as higher level product such as air temperature or comfort index. It will also provide valuable opportunities for other application in urban climate studies, for example the validation of microclimate models. This presentation aims at presenting the various data acquired during the CAMCATT-AI4GEO field experiment in relation to the foreseen objectives and potential future applications.

In 2023, Satellite Vu will launch its first constellation of satellites acquiring high-resolution Mid Wave Infrared imagery and video from low Earth orbit. The main specifications of the first satellite are thermal resolution of 3.5m at Nadir, field of view of 3.5 x 4.4 km, off-pointing ±45 degrees, and thermal sensitivity <2K “ 300k. The imagery is produced in the 3.7-5.0um MWIR waveband. This opens new perspectives and application possibilities, including precision agriculture, monitoring the thermal efficiency of buildings, the effect of the urban heat island effect, improving maritime surveillance and tracking emergency situations including wildfires. Within the earth observation domain, thermal intelligence is mainly acquired with the 100m Sentinel-2. Satellite Vu’s Hot-Sat will provide utility in mapping more granular patterns, detecting hotspots and identifying activity at a level unseen before. The novel technology will provide the capability to differentiate between objects and surfaces of different temperatures and emissivity. Monitoring the urban heat island effect requires the operational ability of the sensor whenever temperatures rise to the levels qualifying for ‘extreme heat’. The potential to access that data increases with the high temporal resolution of HotSat-1, varying between 3-5 days. Satellite Vu aims to provide a service on an international scale thanks to increasing awareness of the importance of collecting thermal data and using it to support public administration (ex. IRIDE EOS4LPA tender). Hot-Sat will be able to assess water quality and report events of thermal pollution. Additionally, it will allow the identification of water reservoirs under the tree canopy or otherwise unidentifiable in the optical spectrum. The imager will capture 25 frames per second (fps), generating up to 60s of video for a point of interest at extreme roll angles. This functionality will be particularly useful to assist with disaster support activities for wildfires, volcanic eruptions and flooding. Key advantages include tracking movement and speed measurement. The presentation will report Satellite Vu constellation capabilities, demonstrate simulated data, and explore how high-resolution satellite imaging will improve performance in high-value applications.

Land Surface Temperature (LST) is an important component of climate and biology, impacting species and ecosystems on a local to a global scale, and is a function of space and time. With the advent of urbanization, more natural land is subjected to the grip of the impervious surface. This has led to a change in the LST values within an urban region and the associated phenomenon of the urban heat island (UHI). Kolkata is the financial and commercial hub of north-eastern India with huge population pressure. This study attempts to understand the Spatio-temporal LST trend over Kolkata and its association with urbanization. The MODIS daily LST (day and night) and Land Cover Land Use (LULC) datasets are used in this study. The data is freely available in the Google Earth Engine (GEE) platform, and only the best quality pixels are considered for further analysis. The study period (March of 2000 to February of 2022) has been segregated into three time periods, annual, seasonal, and monthly, with each being further grouped into day and night. The seasons are classified as pre-monsoon, monsoon, post-monsoon, and winter. The trend in the LST is assessed by the non-parametric Modified Mann Kendall (MMK) test at a significance level of 0.05. The Theil-Sen’s slope is estimated to quantify the magnitude of the trend. The result shows that Kolkata has been warming over the years during the day. The winters are getting colder for some regions of the city, and the monsoon and the post-monsoon are getting warmer for the majority portion of the city. However, there is no significant trend associated with the night-time annual LST. The night-time LST of the pre-monsoon season has an increasing trend. To quantify the contribution of urban land to the urban heat island (UHI) phenomenon, the Urban Heat Island Ratio Index (URI) is calculated. URI is growing both day and night, but very slowly on average.

Near Surface Air Temperature (NSAT) is an important meteorological quantity vastly used in many fields such as agriculture, environmental monitoring, or societal health. NSAT is physically measured 2 m above ground by sensors at meteorological stations. Such measurements represent only a limited surrounding area. Further, available meteorological stations are limited in number with a sparse and non-uniform distribution around the globe. Given the underlying surface heterogeneity at different meteorological stations, it is difficult to create continuous maps from in-situ measurements using interpolation techniques. This brings remote sensing in demand to provide an efficient alternative. Over decades, Land Surface Temperature (LST) has been retrieved using Thermal infrared (TIR) sensor measurements. Using LST as an input, researchers studied different methodologies dedicated to specific regions of the country to retrieve NSAT. These studies confirm a correlation between LST and NSAT. However, available research studies commonly lack generality as all of them are restricted to specific regions or countries and few also depend strongly on auxiliary data input. In addition, a direct end-to-end relationship between TIR measurements and NSAT has not yet been confirmed. These limitations motivated our study to investigate the relationship between TIR measurements and NSAT on a global scale with generality in mind. Our study uses the Landsat-8 bands 10 and 11 with thermal infrared wavelengths around 11 and 12 micrometers along with two different sources of NSAT derived from the Global Historical Climatology Network version 3 (GHCN-v3) data and ERA5-Land data. The GHCN-v3 data set provides in-situ measurements at local weather stations in the form of daily resolution, whereas the ERA5-Land data set provides hourly resolution air temperature generated by a numerical meteorological model. Using the high spatial resolution of the ERA5-Land data set and to ensure representativeness over all continents, climate regions, and land cover types, we sample ground coordinates in a pseudo-random, stratified manner, using the Köppen Geiger climate classification map and Copernicus Global Land Cover layers (CGLS) as guidance. The study uses linear regression, Random Sample Consensus, and a Multi-Layer Perceptron (MLP) relating TIR measurements and NSAT values. MLP provides the best results by a wide margin, indicating that – while a relationship between TIR and NSAT values certainly does exist – it is non-linear rather than of linear nature.

Cities are generally warmer than their surroundings. This phenomenon is known as the Urban Heat Island (UHI) and is one of the clearest examples of human-induced climate modification. UHIs increase the cooling energy demand, aggravate the feeling of thermal discomfort, and influence air quality. As such, they impact the health and welfare of the urban population and increase the carbon footprint of cities. The most commonly studied UHIs are the canopy layer (CUHI) and surface (SUHI) heat islands, which differ fundamentally in their energetic basis and temporal characteristics. SUHIs result from modifications of the surface energy balance at urban facets, canyons, and neighborhoods and are usually estimated from remotely sensed Land Surface Temperature (LST) data. The study of SUHIs using satellite remote sensing has attracted considerable attention in the last two decades, however the published literature is full of contradicting results and erroneous conclusions. One of the reasons for this is the lack of datasets that provide estimates of the SUHI intensity that are consistent between cities and climate zones. In this work we present a global dataset that provides such estimates and describe (i) the method used for delineating cities across the globe in a way that is consistent and suitable for the study of the urban thermal environment; (ii) the method for quality-filtering and processing the LST data; and (iii) the method for estimating the total uncertainty of each SUHII estimate. The developed dataset is based on 20 years of LST data from the Moderate Resolution Imaging Spectroradiometer (MODIS) derived from the MOD11A1, MYD11A1, and MYD21A1N collection 6.1 data products and covers more than 1500 cities around the globe.

The urban thermal environment is an important aspect of evaluating urban ecological environment and land surface temperature (LST) in urban areas is one of the main indicators to reveal this. NASA ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) was launched in 2018 and provided surface temperature data at a spatial resolution of 70 m. In this study, we apply LST data from ECOSTRESS to evaluate different thermal indices in urban areas: the Surface Urban Heat Island (SUHI), the Urban Thermal Field Variance Index (UTFVI) and the Discomfort Index (DI). The indices were estimated in the city of Valencia, a Mediterranean city with flat terrain. The result shows the great potential of ECOSTRESS for understanding the urban thermal environment and lays the groundwork for future high-resolution thermal missions to be launched in the current decade that will help urban planning and the formulation of heat mitigation strategies.

Due to ongoing climate change and urbanization, societies face challenges concerning environmental quality, energy management and citizens’ health. While many past observational and modelling studies concentrated on understanding urban microclimate and how humans experience this, focus has been on relatively modern infrastructure (“street canyons”) regarding modelling and observational efforts which showed less success over historical districts. Many cities have a significant share of aged and historical buildings with unique and different street profiles from modern infrastructure, which raises additional challenges in the energy transition due to low energy-efficiency and restrictions to required interventions. Our research programme will develop a high-tech sensing and design system aiming at detection, reduction and prevention (by monitoring and design) of heat-stress occurring due to ageing of built environmental settings and buildings in cities, through socio-technical solutions. This integral system will detect and forecast spatiotemporal patterns of heat stress at unprecedented resolutions (1m scale), aiming at technological solutions to reduce and mitigate indoor and outdoor heat stress through developing urban design guidelines and connecting the energy transition, housing demands, repurposing areas, climate adaptation and digitalisation. The HERITAGE high-tech sensing and design system necessitates a multi-disciplinary research ecosystem approach involving earth observation, urban hydro-meteorology and climatology, urban design and sustainable infrastructural energy systems; i.e. expertise-fields well represented by the consortium. Therefore, parallel to the sensing, long-term research lines are rolled out on robust hydro-meteorological, design and energy solutions, both (sensing and technological solutions) at multiple spatiotemporal scales and forms. Concretely, these research lines fill knowledge gaps in climate policies through innovative techniques for analysis, simulation, development and experimental testing of newly designed (1) multiscale urban heritage canopy layer schemes for climate models (2) multiscale form-microclimatic relationships and (3) sustainable energy systems, all ideally suited for application in aged neighbourhoods and buildings. Keywords: Urban heat, remote sensing, spatiotemporal modelling, building energy system, urban design

Temperature-emissivity separation (TES) is the choosen method for TRISHNA project to retrieve the two Essential Climate Variable (ECV) from InfraRed Themal (IRT) data, that are Land Surface Emissivity (LSE) and Land Surface Temperature (LST). LSE determination and retrieval in particular, remains a critical issue. Therefore, improved determination of the spectral emissivity is mandatory for a full exploitation of future TRISHNA data sets to estimate a reliable LST product with sub-degree absolute accuracy. In this work we evaluated different candidate methods, notably from statistical relationships between LSE and the corresponding reflectances from VIS-NIR channels. The present work was performed from measured and simulated spectral databases covering the visible and thermal infrared spectra. About 100000 synthetic spectrum where generated in order to cover the best canopy variability range and where used to derive sufficiently robust and generic empirical relationships at canopy scale between VIS-NIR reflectances (THRISNA Channels) and spectral LSE (THRISNA Channels). The work focused on natural cover, and is not extrapolable to artificial cover. The best results were obtained using neural network approaches with rmse on LSE of 0.005, 0.006, 0.003, 0.003 for respectively the IRT channels 8, 9, 10 and 11of TRISHNA. We attempted to validate the relationships in the context of the TRISHNA and LANDSAT spectral bands using data from the AHSPECT-2015 hyperspectral and high resolution airborne campaign. Hyperspectral data were converted into TRISHNA-like by spectral responses convolution then we have compared LST estimated from single channel approach by use of LSE estimated from the relationships and LST estimated from the TES processing chain in development for TRISHNA. Both methods requires a good correction of atmospheric effect performed by MODTRAN simulations. Different LSE obtained were evaluated and mean difference obtained between LST from both approach is about 0.5K with wariations between land cover surfaces type. Provided emissivity estimates are deemed reliable, we will introduce them into the TES as fist guess with the goal to help the method to converge.

Thermal satellite imagery has the potential to provide valuable information on the thermal behavior of the most active volcanoes worldwide, which is helpful in reducing the extensive field efforts. With the rapid development of Earth observation technology, multiple thermal sensors are currently in orbit offering coverage of even for remote and inaccessible volcanoes at a frequency of at least once per day. These sensors provide data over potentially hazardous, high-temperature phenomena, such as volcanic eruptions, with relatively low cost and no risk to the end user. However, the data volume produced by satellite thermal sensors is too large to be manually processed and analyzed on a daily basis and global scale. In recent years, artificial intelligence techniques are the fastest-growing trend in remote sensing data analysis applications, which often rely on massive archive of data. Here, we present an approach to detect volcanic thermal anomalies, exploiting the advantages of a SqueezeNet-CNN already trained with the ImageNet dataset, applying the transfer learning technique. We downloaded the SqueezeNet model and we applied the transfer leaning technique to readjust the parameters of the model, using a new image dataset for volcanic thermal activity, defining two classes, i.e., presence of thermal anomalies (lava flows and intracrateric activity), and absence of thermal anomalies (volcanoes at rest). We demonstrate our approach to study the thermal features at several active volcanoes around the world, using the thermal images acquired by the high spatial resolution ESA Sentinel-2 and NASA Landsat 8 satellites (from 10 m to 60 m). The satellite images were obtained via Google Earth Engine, a cloud-based platform for satellite data analysis. The SqueezeNet model was developed in Google Colaboratory, a cloud platform hosted by Google to program in Python, for data analysis, statistics, machine learning, and deep learning applications. Also comparing this model with other similar approaches, this one returns good performances, making possible to detect the presence of volcanic thermal anomaly with an accuracy of 97%. This system is applicable to all the active volcanoes around the world, alerting in case of volcanic unrest.

ESA is currently planning the development of six new missions that may form part of the expansion of the Space Component of the European Union Copernicus Program. One of these missions is a Land Surface Temperature Monitoring (LSTM) mission to support agriculture, hydrology and food security. Land Surface Temperature (LST) is a basic determinant of the terrestrial thermal behaviour, as it controls the effective radiating temperature of the Earth’s surface. It is influenced by land/atmosphere boundary conditions and exerts control over the partitioning of energy into latent and sensible heat fluxes and the heat flux into the ground. LST is distinct from air temperature, with differences between concurrent measurements of the same scene being as a much as 20 K. Furthermore, it displays a strong diurnal cycle and is sensitive to surface characteristics such as vegetation cover, and soil moisture. LSTM provides the opportunity to enhance an expand our knowledge in these key areas and as such requires scientific study throughout its development to ensure the eventual outputs. This study has the primary aim of supporting the development of LSTM from a scientific basis through the use of advanced simulation of the various instrument effects on the performance of the LSTM mission. Providing updates and analysis of the MRD requirements, whilst providing a review of the methodologies and practises of the OPSI activities relating to the scientific performance

In preparation for the future satellite mission TRISHNA – which should be launched in 2025- there are several on-going studies underway in order to fully exploit the anticipated data that will be produced. TRISHNA will be the first mission to collect both visible and thermal images with a high resolution (around 57m at nadir) and high revisit (3 days) with global coverage. In order to obtain such characteristics, TRISHNA will have a large viewing angle (35°) and different viewing angles during 8 day periods. These characteristics, coupled with the high-resolution, high-revisit of TRISHNA thus require preparatory analyses. Such studies range from onboard calibration, atmospheric correction to applied topics such as Land Surface Temperature direction effects, temperature/emissivity separation, turbulence, land use, evaporation, urban climate, coastal and inland water bodes… Each of these topics having their own complexity. They, however, all require a shared validation methodology with traceable calibrations, standardised protocols, data treatment… The aim of this presentation is to provide an overview of the ongoing applied TRISHNA validation topics such that the workshop participants will have vision of the current status of TRISHNA Cal/Val activities. Topics such as onboard calibration and radiometric calibration will be treated elsewhere. The aim being to promote during the workshop discussion, data, sharing, protocols.. between colleagues and the different satellite missions and to identify common research objectives.

The scope of the present work is to define and describe the characteristics of Italian test sites. The sites should be suitable for calibration and validation purposes to support incoming space missions equipped with thermal sensors. In this context, we have selected five volcanic areas monitored by the INGV that is institutionally responsible for the surveillance of Italian volcanoes. Vesuvio volcano and Campi Flegrei caldera are quiescent volcanoes and represent the main volcanic hazard issue in Italy, due to the more likely explosive scenario in case of renewal of eruptive activity and also characterized by a very high risk because located in a highly urbanized area. Vulcano Island is an active volcano showing fumarole fields and gases emissions. Stromboli presents normally regular explosions throwing out glowing lava from several vents inside its summit crater. Mt. Etna, a natural laboratory for its continuous eruptions, lava flows and gases emissions from the summit craters plumes. The five volcanic areas are suitable to validate a large number of satellite products derived from MIR-TIR data. For calibration purposes, the Mt. Etna summit area (Piano dele Concazze) offers a calibration site at 3000 m a.s.l., with homogeneous surface materials (basaltic tephra) and high emissivity values in the range 8-14 µm. A further calibration site has been identified on Stagno Sale Porcus (Sardegna, Italy) which represents an extensive complex of seasonal, saline lagoons that dry out in summer, leaving a hard salt crust. In this work, we show the spectral characteristics of the selected sites surfaces in terms of emissivity using current satellite data (ASTER, ECOSTRESS) and their use as Italian reference sites for volcanic/geological products validation and possibly to support vicarious calibration for next generation of thermal sensors payloads on SBG-TIR, ESA-LSTM and CNS-TRISHNA space missions.

Since the 1990’s, CNES has developed a strong experience in automated instrumented sites for the calibration of multispectral satellites in a range of wavelengths from 400 to 2500 nm. With the growth of future thermal infrared missions such as Trishna (CNES/ISRO), SBG (NASA) or LSTM (ESA), CNES has the willingness to extend its instrumented sites to the thermal infrared in order to prepare for the in-orbit CAL/VAL of these coming instruments. For this purpose, it has been decided to deploy a new 9 m mast in the La Crau site, which is already a member of RADCALNET, to install and analyze data from several thermal infrared instruments. The NASA JPL designed an automated robust broadband thermal radiometer which acquires day and night measurements in four different zenithal angles and with an internal active blackbody. This instrument has been deployed by JPL for years on various sites, such as Lake Tahoe, Salton Sea and Russell Ranch. The principle is to adjust the blackbody temperature to match the target and the blackbody measurements made by the instrument. Therefore, this radiometer provides surface brightness temperature in one spectral band. CIMEL, a French company, markets the CE 312, which acquires measurements in five narrow spectral bands and one broad band. Emissivity and temperature of the surface can then be retrieved simultaneously by applying a temperature and emissivity separation algorithm on the data. However, this system is not robust for autonomous acquisitions on the long term and requires some adjustments to be automatic and deployed on a mast in La Crau. These instruments will allow calibrating Level-1 and validating Level-2 products. The main characteristics and measurement protocols, and the status of their installation in the La Crau site will be described as well as the processing applied on the measurements. The first results will be discussed and analyzed regarding the final in-orbit calibration objective.

The calibration of spaceborne thermal imaging remote sensing instruments has its own unique set of challenges. A comprehensive pre-launch calibration is fundamental to successful on-orbit operations. A variety of well-understood techniques are employed on orbit to monitor instrument performance and maintain calibration within a defined uncertainty including techniques to intercompare data between multiple sensors. High-quality intercomparisons require understanding the radiometric, geometric, and spectral calibrations and associated uncertainties of the various sensors. Traditional vicarious methods of sensor intercomparison using observations of specific Earth surface targets, while relatively straightforward for visible and shortwave infrared sensors, is more challenging for thermal infrared sensors due to effects from spectral emissivity and temporal variations in atmospheric conditions and surface emissivity and temperature. The current work presents an overview of the prelaunch calibration of the two TIRS instruments that are currently on orbit as part of the Landsat 8 and 9 missions. The radiometric calibration discussion concentrates on the SI-traceability of the absolute radiometric calibration and its transfer to orbit using both an onboard blackbody as well as vicarious methods. The geometric calibration of the sensors is also described as well as a description of the improvements made to Landsat-9 TIRS to mitigate a stray light issue found in Landsat-8 TIRS data. On-orbit validation of the spatial quality of TIRS is described as well as radiometric intercomparisons between the two TIRS instruments concentrating on data from an underfly of Landsat 8 by Landsat 9 as it was raised to its final orbit. Lessons learned from the TIRS case are given and how they can be applied to future missions such as LSTM, TRISHNA, and SBG including expected challenges when attempting to harmonize data from systems operating in differing orbits, spectral bandpasses, and mission operation lifetimes.

By 2030, there should be three new high resolution thermal infra-red satellite imaging mission operation in space: Trishna, LSTM and SBG. All those missions have similar spatial resolutions of 60 meters or better at Nadir and a large field of view (fov), exceeding 30°. In the case of Trishna, a given location will be observed 3 times in 8 days, each time with a different viewing angle. Even if LSTM and SBG have constant view angles, merging time-series from the three sensors will require quantifying and modelling directional effects at high resolution. While the traditional approach of modelling such effects is to gather in situ data and use them together with modelling tools such as DART, in this work we try an orthogonal approach by looking for directional effects into archives of already flying missions in the target resolution range: Landsat-8 (Collection 2, level 2), Ecostress (Collection 1, level 1B and 2), the NASA sensor flying on the ISS (fov ± 26°) and Master (https://masterprojects.jpl.nasa.gov/, level 1B and 2), a NASA airborne sensor (fov ± 45°). We searched for pairs of images that have been acquired over the same location within a 10 minutes time frame. We found 205 pairs of nearly simultaneous Ecostress and Landsat-8 images, for which we can compare LST as well as emissivities and radiances, for a total of 24.26M of pixels. We have observed an LST bias of 2.31 K (Landsat-8 being hotter that Ecostress) with a standard deviation of 1.56 K. Nearly 60% of the bias and 30% of the standard deviation can be explained by difference in emissivities and around 30% of the bias and 20% of the standard deviation can be explained by local observation time deltas between Landsat-8 and Ecostress pixels, which shows that Landsat-8 and Ecostress measurement are well inter-calibrated. At global scale, there is no significant directional effect within the ± 26° of Ecostress field of view. Moving on to the analysis of Master matches with Landsat-8, we found 19 valid matches for which the individual analysis shows clear trends related to viewing and solar angles. We can observe that the LST difference exhibits a parabola shape, temperature being cooler than Landsat-8 for high viewing angles, except when coming close to the hotspot, where we can see that Master tends to be hotter than Landsat-8. This preliminary analysis will be consolidated, quantified and complimented with the analysis of matches between Master and Ecostress and outcomes will be presented at the workshop.

Land surface temperature (LST) was defined as an Essential Climate Variable by the Global Climate Observing System of the World Meteorological Organization and Landsat series allow retrieving LSTs at high spatial resolution. Landsat 9 (L9), a joint mission of NASA and the U.S. Geological Survey, was launched in September 2021 and started operational phase by the end of January 2022. Thermal Infra-Red Sensor (TIRS-2) is on board L9 and acquires TIR data with two spectral bands (bands 10 and 11 centered at 10.8 and 12.0 micrometers) at a spatial resolution of 100 m. Reference ground thermal radiance data measured along transects in a rice crop site (the Valencia Test Site) through 2022 were used to evaluate the calibration of the L9 TIRS-2 bands and the derived LSTs provided in the L2 product, which uses a single-channel correction algorithm for band 10. The site has different land covers due to the crop changes through the year, from null to full vegetation cover, and water surface when flooded, which makes it interesting for calibration/validation (CAL/VAL) activities and allow covering a wide range of LSTs. TIR emissivities were also measured at the site. First calibration results showed that both systematic and random uncertainties were lower than 1 K, with a slight underestimation shown for both L9 TIRS-2 bands. Negligible systematic uncertainties and root-mean-square differences lower than 1.5 K were obtained when evaluating the L2 LST product at the site. The study had the financial support of the projects Tool4Extreme PID2020-118797RBI00 funded by MCIN/AEI/10.13039/501100011033 and PROMETEO/2021/016 funded by Generalitat Valenciana.

The stated goal of NASA’s Earth Science Research Program is to utilize global measurements to understand the Earth system and its interactions as steps toward the prediction of Earth system behavior. NASA has identified the provision of well-calibrated, multiyear and multi-satellite data and product series as a key requirement for meeting this goal. In order to help address this goal we have established two automated validation sites where the necessary measurements for validating mid and thermal infrared data from spaceborne and airborne sensors are made every few minutes on a continuous basis. The two automated validation sites are located at Lake Tahoe CA/NV and Salton Sea CA. The Lake Tahoe site was established in 1999 and the Salton Sea site was established in 2008. Lake Tahoe is ideally suited for validation of mid and thermal infrared data for several reasons including its size, homogeneity, elevation, accessibility and composition. In order to use Lake Tahoe for validation, 4 buoys have been deployed. Each buoy includes a custom-built highly accurate (50mK) radiometer measuring the surface skin temperature and several bulk temperature probes that trail behind the buoy. Each buoy includes a logging system with cellular access and two full meteorological station measuring wind speed, wind direction, relative humidity and net radiation. All the measurements are made every few minutes and downloaded hourly via a cellular modem. The buoy measurements are supplemented with a variety of atmospheric measurements made on-shore. The Salton Sea site was established in 2008 to validate high water temperatures, up to 35 C and evaluate the performance of surface temperature retrieval algorithms under wet and dry atmospheres depending on time of year. Data from the sites have been used to validate numerous satellite instruments from ESA, NASA, and NOAA and plans are underway to utilize the sites to validate measurements from several planned satellite-sensor systems. These include TRISHNA being developed by CNES and ISRO with a launch scheduled in 2025, SBG being developed by NASA and ASI with a launch planned for 2027 and LSTM being developed by ESA with a launch scheduled in 2029. In order to validate these new satellite-sensor systems several new sites are being developed in multiple countries around the world that complement the measurements made a Lake Tahoe and Salton Sea. We will present results using the Lake Tahoe and Salton Sea sites and discuss other potential automated validation sites that could be used in combination to create a validation network for mid and thermal infrared sensor measurements.

Viewing and illumination geometries are known to have significant impacts on remotely sensed land surface temperature (LST) retrieval, particularly in heterogeneous regions. Previous studies have attempted to quantify these impacts mainly through modelling or via analysis of observations collected by aircraft-mounted single-band thermal imagers. However, models necessarily simplify real-world conditions, whilst single band thermal imagers have limitations such as their radiometric accuracy, the fact their data cannot be used in so-called Temperature-Emissivity-Separation (TES) algorithms, and the fact that they cannot for example be applied to consider the angular variability of emissivity. A joint ESA-NASA funded airborne and field campaign in Italy and France will be conducted to further investigate these issues, building on the SwathSense 2021 campaign. A central and rather unique component, aimed at reducing past limitations in the collected datasets, will be the concurrent acquisition of LWIR hyperspectral data at both nadir and off-nadir viewing angles. This is considered extremely beneficial as it overcomes a key challenge limiting the analysis of nadir- and off-nadir LWIR data collected non-simultaneously (e.g. from one sensor on neighbouring flightlines), namely the challenge of disaggregating the sought-after angular effects from the real surface temperature changes that may have occurred over the measurement interval. The hyperspectral sensors deployed will be NASA-JPL’s LWIR state-of-the-art Hyperspectral Thermal Emission Spectrometer (HyTES) and the National Centre for Earth Observation’s Airborne Earth Observatory (NAEO)’s modified commercial LWIR imager (AisaOWL), which is mounted such that it can move from viewing at nadir to viewing at beyond 45°. Data will be collected over swaths that enable data simulation at the satellite scale to be performed, with the aim of understanding and potentially developing adjustments for wide view angle satellite-based LST retrievals. When combined with other airborne instrumentation to be flown (including a FENIX 1K visible and shortwave infrared hyperspectral imager), and the ground instrumentation to be deployed, these datasets also have the potential to enable evaluation of angular effects on remotely sensed evapotranspiration estimates. The presentation will provide a summary of the findings of SwathSense 2021, detail the developments made since then, and provide an overview of the new campaign plans and how it builds on prior work with a view to being highly relevant to future satellite thermal missions.

Agriculture is the largest consumer of water worldwide, accounting for about 70% of the global freshwater      withdrawals. Thus, crop water use efficiency and impacts of water stress on crop water consumption are the key concerns for agricultural water management. Present study investigates the variability of evapotranspiration (ET) and crop water use efficiency by integrating very high spatial resolution thermal infrared (TIR) data from airborne measurements and visible to near infrared data from Planet satellite with a numerical water-energy balance model and a diagnostic surface energy balance model.      The analysis is done for an agriculture area in central Italy near the city of Modena, where several fruit trees fields are present along with fresh vegetables. An intensive airborne campaign was organized in the summer of 2022 for three consecutive days in July. A hyperspectral TIR camera (Telops Hyper-Cam LW) has been operated at a spectral resolution of 8 cm-1, resulting in 64 wavebands, and covering a wavelength range of 850 cm-1 to 1350 cm-1 (7,39 µm – 11,8 µm). During the 3 days of flight acquisitions, three overpasses per day were planned: 9:00, 12:30 and 16:00 h, at both 4 and 1 m spatial resolution. Planet data at 3.7 m spatial resolution were used to derive different vegetation indices, such as vegetation fraction coverage and leaf area index. During airborne overpasses ground data of spectral reflectance, vegetation variables, LST and soil water content (SM) were collected in different fields. In addition, two fields were monitored with an eddy covariance station and SM profile measurements. To investigate the diurnal and spatial patterns of ET, SM variability and crop water use efficiency, we used two numerical models: the surface energy balance model STIC based on Penman-Monteith and Shuttleworth-Wallace (Mallick et al., 2018) and the water-energy balance model FEST-EWB which computes continuously in time and is distributed in space soil moisture and evapotranspiration fluxes solving for a land surface temperature that closes the energy–water balance equations (Corbari et al., 2011). Traditional residual energy balance models (SEBS (Su et al., 2002) and TSEB (Norman and Kustas, 1995)) are also employed. Differences and similarities in ET estimates have been analysed from the models for different SM conditions and crop types, considering crop water use efficiency and water stress, and have been compared to eddy covariance measurements for accuracy evaluation considering both instantaneous and daily data.