Conference Agenda

Land Surface Temperature (LST) is one of the main quantities governing the energy exchange between the surface and atmosphere. This abstract provides a summary of the latest progress of Dragon-5 project (59318), including (1) all-weather LST generating methods and its implementation; (2) all-weather LST downscaling methods; and (3) satellite LST validation against in-situ LST.

To investigate the temporal and spatial variations of LST in China, long-term, high-quality, and all-weather LST datasets are urgently needed. However, the publicly reported all-weather LSTs are not available during the temporal gaps of MODIS between 2000 and 2002. Therefore, the enhanced RTM (E-RTM) method was proposed to produce a daily 1-km all-weather LST dataset for the Chinese landmass and surrounding areas, i.e. TRIMS LST (Tang et al., 2023). Validation against in-situ LST shows a MBE range of -2.26~1.73 K and a RMSE range of 0.80~3.68 K, with slightly better accuracy than the MODIS. The TRIMS LST has already been used by scientific communities in various applications, e.g. evapotranspiration estimation, and urban heat island (UHI) modelling.

A method integrating reanalysis data and TIR data from geostationary satellites (RTG) was proposed for reconstructing hourly all-weather LST (Ding et al., 2022). The method was implemented over Tibetan Plateau with the Chinese Fengyun-4A (FY-4A) TIR LST and China Land Surface Data Assimilation System (CLDAS) data. Validation against in-situ LST shows that the accuracy of the all-weather LST is better than FY-4A LST and CLDAS LST. The mean RMSE is better than 3.94 K for all conditions, respectively. The reconstructed all-weather LST also has good image quality and provides reliable spatial patterns.

To obtain high spatial resolution all-weather LST, two downscaling methods were developed. The first is downscaling TRIMS LST using LightGBM from 1 km to 250 m over Southeast Tibet (Huang et al., 2021). Validation against in-situ temperature shows a MBE of 0.74 K (-0.01 K) and RMSE of 2.25 K (2.15 K) for daytime and nighttime, respectively. The second is a method, which is proposed based on the geographically weighted regression (GWR) and random forest (RF), and considering the weights of LST descriptors (Ding et al., 2023). The method was tested to downscale 1000-m aggregated ASTER LST to 100 m. Validation against in-situ LST shows that the MBE and RMSE can be reduced more than 0.22 K and 0.1 K in Beijing and Zhangye. The explorations in downscaling provide the basis for obtaining high-resolution all-weather LST.

To validate the satellite-retrieved LST at the kilometer scale, we proposed a temporal variation method for evaluating the ground station’s spatial representativeness (Ma et al., 2021). The spatial representativeness indicator is defined as the LST difference between the in-situ radiometer’s FOV and satellite pixel scale and extended in temporal with the temporal variation of LST and related parameters. Then, the in-situ LST was convert to pixel scale to validate the MODIS and AATSR LST. Results show that, among the selected stations, a systematic bias of -1.95~5.60 K and a random error of 0.07~3.72 K can be found for the validation results if the station’s spatial representativeness is ignored. Therefore, it is suggested that the ground station’s spatial representativeness over inhomogeneous surfaces should be considered in LST validation, as well as other related parameters. In further research, the spatial representativeness of KIT’s station and HiWATER station will be evaluated and then used in all-weather LST validation.

Since 2008 KIT operates a permanent LST validation station near Gobabeb, Namibia (Göttsche et al., 2022). In the rainy season of 2010/2011 the largest amount of rainfall in recorded history was measured at Gobabeb’s meteorogical station (Eckhardt et al., 2013), which resulted in an exceptionally strong growth of grass over large parts of the gravel plains. Due to the extreme atmospheric conditions and the changes in biophysical surface properties, LST retrievals for this period can provide interesting insights into the performance of LST products. Here, two all-weather LST products are compared over the gravel plains: 1) all-weather LST obtained with Reanalysis and Thermal infrared remote sensing Merging (RTM) (Zhang et al., 2021), which uses reanalysis and Moderate Resolution Imaging Spectroradiometer (MODIS) thermal data to estimate LSTs at cloudy MODIS overpasses, and 2) the operational all-weather LST product of the Land Surface Analysis (LSA) Satellite Application Facility (SAF), which merges clear-sky MSG/SEVIRI LST with surface temperatures from a Soil-Vegetation-Atmosphere (SVAT) model driven by LSA SAF satellite products (Martins et al., 2019). The two all-weather LST products are validated with in-situ LST and their spatial variation over the gravel plains is investigated, thereby providing a comprehensive analysis of their performance over a broad range of atmospheric and surface conditions encountered at Gobabeb.

Eckardt, F.D., Soderberg, K., Coop, L.J., et al., 2013. The nature of moisture at Gobabeb, in the central Namib Desert. J ARID ENVIRON, 93, 7–19.

Ding, L., Zhou, J., Li, Z.-L., et al., 2022. Reconstruction of Hourly All-Weather Land Surface Temperature by Integrating Reanalysis Data and Thermal Infrared Data From Geostationary Satellites (RTG). IEEE Trans. Geosci. Remote Sensing, 60, 1–17.

Ding, L., Zhou, J., Ma, J., et al., 2023. A Spatial Downscaling Approach for Land Surface Temperature by Considering Descriptor Weight. IEEE Geosci. Remote Sensing Lett., 20, 1–5.

Göttsche, F.-M., Cermak, J., Marais, E., et al., 2022. Validation of Satellite-Retrieved Land SurfaceTemperature (LST) Products at Gobabeb, Namibia. Journal Namibia Scientific Society, 69, 43–61.

Huang Z., Zhou J, Ding L., et al., 2021. Toward the method for generating 250-m all-weather land surface temperature for glacier regions in Southeast Tibet. Journal of Remote Sensing, 25, 1873–1888.

Ma, J., Zhou, J., Liu, S., et al., 2021. Continuous evaluation of the spatial representativeness of land surface temperature validation sites. Remote Sensing of Environment, 265, 112669.

Martins, J. P. A., Trigo, I. F., Ghilain, N., et al., 2019. An All-Weather Land Surface Temperature Product Based on MSG/SEVIRI Observations. Remote Sensing, 11(24), 3044.

Tang, W., Zhou, J., Ma, J., et al., 2023. TRIMS LST: A daily 1-km all-weather land surface temperature dataset for the Chinese landmass and surrounding areas (2000–2021), Earth Syst. Sci. Data Discuss. in review.

.