Comparison of gap-filling methods for producing all-weather daily remotely sensed near-surface air temperature

Abstract

Thermal infrared remotely sensed near-surface air temperature (Ta) can provide gridded temperature information at relatively high spatial and temporal resolutions, and such measurements are thus widely used as an essential environmental parameter in numerous fields. However, data gaps caused by clouds highly restrict the applicability of remotely sensed Ta. Only a few studies have explored the production of seamless remotely sensed Ta products, and all of them estimated Ta from the gap-filled land surface temperature (LST). This study first estimates the daily minimum, average and maximum Ta of clear-sky pixels from remote sensing data over the Yangtze River Delta (YRD) and the Ningxia Autonomous Region (NAR), China, during 2016–2020, and then applies five gap-filling methods, including spatial, temporal, spatiotemporal and two multisource fusion-based gap-filling methods, to fill the data gaps in the remotely sensed Ta data. The performances of these methods under different cloud, terrain and landscape conditions are also assessed. The validation results indicate that the Temporal Fourier analysis (TFA) method exhibits high accuracy, good robustness under various cloud and surface conditions, and good ability to describe spatial details of Ta of cloud-cover areas. It is the most suitable method to fill data gaps in remotely sensed Ta. This study provides a valuable reference for selecting appropriate methods to develop seamless remotely sensed Ta products.