The land surface temperature (LST) can be estimated from satellite passive microwave observations. Compared with the infrared satellite retrievals, the microwave retrievals suffer from very limited cloud contamination, although coarser spatial resolutions and a stronger dependence on a more varying surface emissivity result in more challenging retrievals. With a ~60% cloud cover in average over the globe, the importance of a long record of “all weather” LST has been recognised, and the ESA Land Surface Temperature Climate Change Initiative project (LST-CCI, http://cci.esa.int/lst) will be providing such a record from the Special Sensor Microwave/Imagers (SSM/I) and the Special Sensor Microwave Imager Sounder (SSMIS) instruments to complement its “clear-sky” infrared records.

The SSM/I and SSMIS LST data record will be based on a simple yet accurate methodology developed to derive LST from microwave conical scanner observations, with the help of pre-calculated land surface microwave emissivities [1], providing 25 years (1992-2017) of twice-per-day (~6.30 am/pm) global LST estimates at a spatial resolution of ~25 km. Initial analyses suggest a theoretical precision of ~2.5 K, although this figure can degrade for challenging retrieval situations, such as snowed regions or dry places with large microwave emission from below the surface. First results from SSM/I retrievals have been compared to infrared estimates from International Satellite Cloud Climatology Project (ISCCP) and from Advanced Along Track Scanning Radiometer (AATSR), under clear-sky conditions. Limited biases are observed (~0.5 K for both comparisons) with a root-mean-square difference (RMSD) of ~5 K, to be compared to the RMSE of ~3.5 K between ISCCP and AATSR.

This methodology can also be applied to other microwave conical scanners. During the ESA GlobTemperature project (http://www.globtemperature.info) three years (2007-2010) of LST from the Earth observation satellite (EOS) Advanced Microwave Scanning Radiometer (AMSR-E) were produced [2,3]. Given the different observation times (~1.30 am/pm) and finer spatial resolution (~12 km), a longer data record from AMSR-like instruments could nicely complement the SSM/I and SSMIS LST estimates. In the future, finer spatial resolutions could be achieved from an instrument such as the Copernicus Imaging Microwave Radiometer (CIMR) [4], which will open the possibility of real time “all weather” LST estimates with resolutions close to 5 km.

[1] Prigent, C., C. Jimenez, and F. Aires, F., Towards ’all weather’, long record, and real-time land surface temperature retrievals from microwave satellite observations, J. Geophys. Res., doi:10.1002/2015JD024402, 2016.

[2] Jimenez, C., C. Prigent, S.L. Ermida, and J-L. Moncet Inversion of AMSR-E observations for land surface temperature estimation - Part 1: Methodology and evaluation with station temperature, J. Geophys. Res., doi:10.1002/2016JD026144, 2017.

[3] Ermida, S.L., C. Jimenez, C., C. Prigent, I.F. Trigo, and C.C DaCamara, Inversion of AMSR-E observations for land surface temperature estimation - Part 2: Global comparison with infrared satellite temperature, J. Geophys. Res., doi:10.1002/2016JD026144, 2017.

[4] Kilic, L., C. Prigent, F. Aires, J. Boutin, G. Heygster, T. T. Tonboe, H. Roquet, C. Jimenez, and C. Donlon, Expected performances of the Copernicus Imaging Microwave Radiometer (CIMR) for an all-weather and high spatial resolution estimation of ocean and sea ice parameters J. Geophys. Res., doi:10.1029/2018JC014408, 2018.

In 2018, the US National Research Council released the second Earth Science Decadal Survey (ESDS). The second survey recommended launching a mission with similar specifications to the Hyperspectral Infrared Imager (HyspIRI) mission recommended in the first ESDS, referred to as the Surface Biology and Geology (SBG) mission. The second ESDS identified SBG as a global mission designed to address a critical set of questions on how the surface of the earth is responding to natural and human-induced changes. The SBG mission requires hyperspectral measurements between 0.38 and 2.5 µm at a spatial scale of 30 m with a minimum revisit of 16-days and a thermal infrared (TIR) multispectral measurements between 4 and 12 µm at a spatial scale of 60 m with a minimum revisit of 5-days. As part of the risk reduction activities for the TIR instrument for HyspIRI a space-ready Prototype HyspIRI Thermal Infrared Radiometer (PHyTIR) was developed in the laboratory for hardware risk reduction and an airborne Hyperspectral Thermal Emission Spectrometer (HyTES) for science risk reduction. In July 2014, the ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) was selected as part of the NASA Earth Ventures Instrument program and launched in June 2018. ECOSTRESS is based on PHyTIR and is being used to address critical science questions related to evapotranspiration. ECOSTRESS has 5 TIR spectral bands, a spatial resolution of 68m x 38m (crosstrack x downtrack) and a revisit of every few days at varying times of day from the International Space Station. HyTES is an airborne instrument with 256 spectral bands in the TIR and pixel sizes of 2-30 m depending on flight altitude. HyTES is one of the few hyperspectral thermal infrared imaging spectrometers. Recently HyTES was modified to fly on the NASA ER2 and completed a campaign in Hawaii, it also flies on a Twin Otter. This presentation will describe the current status of these activities with an emphasis on the derivation of land surface temperature from the different sensors.

The Chinese high resolution (Geofen-5, GF-5) satellite provides global land surface observation with four thermal infrared channels from 8.0 um to 12.0 um at 40m, and its land surface temperature (LST) and emissivity products will be used in agricultural drought and urban heat island monitoring, mineral exploration, surface energy budget estimate and so on. The authors developed several new algorithms for GF-5 LST and emissivity retrieval:
(1) A Four-channel split-window algorithm that fully used the four TIR channel and was found to be better than the conventional two-channel and three-channel split-window algorithms;
(2) A local hybrid algorithm that combines the split-window algorithm and TES (temperature-emissivity separation) algorithm as: the emissivity calculated from the TES algorithm can be used in the SW algorithm, and the LST from the SW algorithm can then be applied to the TES algorithm as an initial value to refine emissivity and LST. Results showed that the hybrid algorithm was less sensitive to the instrument noise and atmospheric data error, and can obtain LST and emissivity with an error less than 1.0 K and 0.015, respectively, better than the single TES or SW algorithm.
(3) A global hybrid algorithm for synchronously retrieving LST without having to input atmospheric and in-prior emissivity data. The process involves the combination of the SW and TES algorithms: the SW algorithm is used to estimate surface-emitting radiance in adjacent TIR bands, and the band surface-emitting radiance is applied to the TES algorithm to retrieve LST and emissivity data. Result showed that the new algorithm can estimate LST and emissivity with errors in the 0.5–1.5 K and 0.007–0.02 range, respectively. Moreover, the LST error of the Hybrid algorithm was equivalent to that of the original TES algorithm with 1%–2% uncertainty involving atmospheric correction of TIR data.
(4) A new algorithm for retrieving urban surface temperature by considering the urban building scattering effect using digital surface model and sky-view factor. Our study indicated that the urban building scattering effect can increase the urban thermal emission up to several Kelvin, and the results from the new algorithm can improve the temperature significantly.
With the above algorithms, we are providing LST and emissivity products from GF-5 observation, and set seven ground-measure sites for the validation purpose.
Moreover, with the support from the Agri-Tech in China UK Newton Network+ (ATCNN) Small Project Award, we have extended those algorithms to retrieve LST from Sentinel-3A SLSTR (Sea and Land Surface Temperature Radiometer) at 1 km scale and used the product to perform agricultural drought monitoring in Hetao plain of Inner Mongolia, China.

Land surface temperature (LST) values retrieved from satellite measurements in the thermal infrared (TIR) may be strongly affected by spatial anisotropy. This effect introduces significant discrepancies among LST estimations from different sensors, overlapping in space and time, that are not related to distinct assumptions in the methodologies of retrieval or different input data used. Furthermore, these directional effects deviate LST products from an ideally defined LST, which should represent the ensemble of directional radiometric temperature of all surface elements within the FOV.
There are two main effects that may contribute to the directional character of LST:
1) Remotely sensed LST in the TIR only indicates the mean temperature of the surfaces that can actually be seen by the sensor. As such, the retrieved LST may change as a result of varying proportions of the different surfaces within the field of view. In this case, LST anisotropy depends not only on structure and location of the surface components relative to each other, but also on the temperature differences among components.
2) At the scale of the individual surface components, directional effects may also arise from the anisotropy of the emissivity. Experimental studies show that for homogeneous surfaces, such as water, bare soils, sand and clay, the emissivity decreases with view zenith angle. However, at the scale of satellite observations, this may not always be the case and there are very few methods that allow retrieving such dependence from satellite data.
In the past few years, the LST requirements have become stricter especially because of climate applications, as LST constitutes an independent temperature data set for quantifying climate change complementary to the near-surface air temperature. Several methodologies have recently been developed within the ESA DUE GlobTemperature project and further explored within the EUMETSAT Satellite Application Facility on Land Surface Analysis (LSA-SAF) with the purpose of better characterizing the impact of angular effects on LST products. These methodologies include: 1) the upscaling of in situ observations taking into account the directionality of LST, allowing a more appropriate validation of satellite retrieved LST; 2) the assessment of the capabilities of different parametric models of directional LST and their respective potential for operational use, in order to provide the LST users with information on expected LST deviations due to directional effects; 3) the adoption of possible strategies to retrieve or model directional effects in emissivity, in particular for desert areas where these effects are expected to be more relevant.
The various methodologies to understand and model the LST directional dependence will be assessed and consolidated with the aim of building up LST Climate Data Records within the framework of the ESA CCI LST project.

Land surface temperature (LST) is a key environmental variable that drives several land surface processes, including radiation budget, heat and water exchanges within surface boundary layer, soil water depletion through evaporation and transpiration, vegetation photosynthesis and soil respiration, pollution degradation, or pathogen dissemination. Vegetation is one of the most important ecosystems of the land surface. If the emissivity or temperature of components was obviously different, the component temperature could be inversed by the multi-angle radiance data. Sentinel-3A data provide the new chance for components temperature retrieval.

We proposed the new analytical directional canopy emissivity model (CE-P) based on spectral invariants from both the perspective of absorption and active emission, then the new directional thermal radiation model for vegetation canopy based on spectral invariants has also been developed. So the Land surface component temperatures of soil-vegetation system can be more accurately estimated based on the new model.

In this study, the frame of component temperature retrieval based on Sentinel-3A data and CE-P model is established. Firstly, the LAI can be retrieved by visible and near infrared Sentinel-3A data, so the multi-angle and multi-band emissivity of canopy can be calculated according to the CE-P model. Then the iterative method combining inverse matrix is chosen to retrieve the soil and vegetation temperature from the directional thermal radiation model using two bands and two angles thermal infrared Sentinel-3A data. Thirdly the local split window method is also specific established to correct the atmospheric effect for Sentinel-3A band. At last the numerical simulation and single and multiple factors sensitivity analysis are carried out to prove the stability and accuracy of components temperature inversion. The row crop was selected as the target to validate the method using Sentinel-3A in Huailai County, Hebei, China, 2018. The feasibility and rationality of this method are validated by the measured temperature obtained by thermal imager in the field test.

Land surface temperature (LST) is the radiant skin temperature of the land surface that can be retrieved from remote sensing instruments. LST has a key variable for Earth System Science as it is the result of various processes involving surface-atmosphere interactions and energy fluxes. We now have the capacity of measuring it accurately across the Earth with decent spatial resolution and adequate revisit frequency for many applications. However, LST can be erroneously interpreted as being equivalent to air temperature. Furthermore, for climate science and policy, temperature is considered to be that which is measured in a weather station under controlled conditions 2 m above the surface. Such measurements of air temperature are typically only available from a limited number of weather stations distributed mainly in developed countries, which in turn may often report time series with missing values, resulting in a record of air temperature observations that is patchy in both space and time. While gridded dataset of air temperature can be generated through spatial-temporal interpolation using process-based models (e.g. climate re-analysis) or geostatistical tools, a more direct approach is to derive it directly from LST. We thus developed a statistical model to predict air temperature from LST globally. The approach is based on establishing statistical regressions between records from weather stations and their corresponding LST measurements. A geographically weighted regression (GWR) is fitted and cross-validated individually for each station in a set of more than 3000 separate stations. However, the relationship between air temperature and LST could even be more consistent over stations with similar climates, despite being separated geographically. To exploit this property, a novel approach called climate space weighted regression (CSWR) was developed in which a similar regression is derived for each station, but instead of basing it on similarity in geographic space, it relies on their similarity in a climate space, defined by the mean annual tem-perature and precipitation regime of the stations. Results from GWR and CSWR are finally combined to make an overall prediction of air temperature using stacked generalization, a method to optimally combine multiple statistical models into an ensemble model. The final results demonstrate higher cross-validation performance than individual models. The method was then used it to generate a novel dataset of monthly air temperature at spatial resolution of 0.05° for the period 2003-2016. We expect this dataset to be useful for various applications involving climate monitoring, Earth system science and land-climate interactions. We further think that the methodology could be adapted to any source of satellite LST data record and by doing so, improve Land Surface Temperature science and increase its reach to a wider community.