Spatial and temporal patterns of drought based on RW-PDSI index on Loess Plateau in the past three decades

Palmer Drought Severity Index (PDSI) is a widely used indicator to assess the intensity of drought based on a similar two-layer bucket (TLB) model, especially in arid and semi-arid regions. However, the TLB model cannot quantify surface water balance, effectively, which may lead to the mis-assessment of drought. Therefore, an optimized PDSI indicator named RW-PDSI based on the water balance is constructed. RW-PDSI is developed using the output of a newly proposed remote-sensing-based water balance assessment tool (RWBAT) model. The RWBAT model is constructed to depict the detailed hydrological processes on a fine spatial scale using highresolution remote sensing data. To test its performance, we collected tree-ring data on the Loess Plateau (LP) to show the drought conditions in past decades (1991-2020). Tree-ring data provide a useful continuous multiyear data archive for drought assessment. Our results showed that the correlation between RW-PDSI and treering data was statistically significant (p < 0.01) at Kongtong and Hengshan sites with 0.65 and 0.64 respectively. Meanwhile, RW-PDSI showed a greater overall correlation with the tree-ring data than the self-calibrating PDSI(SC-PDSI) and the Standardized Precipitation Evapotranspiration Index (SPEI). Results also illustrated that RW-PDSI of the LP region increased in the past three decades, indicating weakening drought frequency and intensity. The area with a significant weakening trend of drought in the investigated region accounts for 71 %. Overall, drought characteristics in LP were well represented by our drought indicators, which could provide research support for the sustainable development of the ecosystem and social economy in drought areas.