Surface Water Availability and Temperature (SWAT): An Innovative Index for Remote Sensing of Drought Observation

Crop water stress and drought status can be indicated through soil moisture (SM). SM-related factors, such as water availability, vegetation state, and temperature, can be inferred by normalized difference latent heat index (NDLI), normalized difference vegetation index (NDVI), and land surface temperature (LST), respectively. This study presents a new surface water availability and temperature (SWAT) index that utilizes remotely sensed data to monitor crop water stress and drought status. The Euclidean distance method was used to analyze the three-dimensional (3-D) scatterplots of NDLI, NDVI, and LST to define the SWAT. Subsequently, the SWAT and two drought indices, including the temperature-vegetation dryness index (TVDI) and temperature-soil moisture dryness index (TMDI), were used to determine the spatial signatures of droughts in Taiwan from 2013 to 2020. These drought signatures were evaluated and compared against corresponding assessments from various indicators, including the in situ SM, TVDI, TMDI, crop water stress index (CWSI), and net primary productivity (NPP). The results show that the TVDI and TMDI were negatively correlated with the in situ SM, with the highest correlation coefficient ( r ) of -0.53 and -0.67, respectively. The SWAT was found to be even more correlated to the SM ( - 0.79 <= r <=- 0.49$ , $p < 0.01$ ) and exhibited more sensitive and stable than other single indices (NDVI, NDLI, and LST) and integrated indices (TVDI and TMDI). According to the spatial observation, the SWAT index was closely related to CWSI ( r = 0.77 , p < 0.01 ). Moreover, the SWAT was in line with NPP ( r = - 0.79 , p < 0.01) and more sensitive to drought than the TVDI and TMDI. Furthermore, the SWAT was highly correlated with the TVDI, TMDI, CWSI, and NPP in most agricultural regions in Taiwan. Overall, the SWAT is proven to be a satellite index effectively describing agricultural drought stress.