Analysis of Slope and NDVI Effects on Land Surface Temperature Retrieval Accuracy in Mountain Area Based on WIS Data of Tiangong-2

Abstract

The relationship between land surface temperature (LST) and Normalized Difference Vegetation Index (NDVI) is evaluated in mountain areas using the Wide-band Imaging Spectrometer (WIS) data on Chinese Tiangong-2 (TG-2) Space Lab. The WIS is composed of three regions: Visible Near-Infrared (VNIR) (14 channels), Short-Wave Infrared (SW) (2 channels), and Thermal Infrared (TIR) (2 channels). The two TIR channels, T1 (8.125–8.825 μm) and T2 (8.925–9.275 μm) have not been fully exploited to extract LST. Thus, a modified split window method was proposed to retrieve LST and evaluated by comparing to the MODIS LST product. Then, the LST retrieval accuracy affected by slope and NDVI is analyzed. Results show that: (a) TG-2 LST is positively correlated to MODIS LST (R2 = 0.74) but overestimated (about 3.5 K); further improvement on the split window method is still needed; (b) the TG-2 LST accuracy is different in vegetation and non-vegetation area, where non-vegetation area with small NDVI standard deviation has better agreement between MODIS LST and TG-2 than vegetation area; (c) the LST retrieval accuracy in vegetation area is independent on slopes; while increasing with the slope in non-vegetation area.