Seasonal and aridity influences on the relationships between drought indices and hydrological variables over China

In recent decades multiple drought indices have been developed to quantify drought characteristics and to examine drought effects. However, little is known about regional applicability of drought indices for monitoring seasonal droughts over China. Here we investigated the capacity of three commonly used drought indices (SPI, SPEI and scPDSI) for showing water deficit as compared to independent precipitation, root zone soil moisture (SM) and terrestrial water storage (TWS) estimates collected from satellite observation and model simulation, by using trend and correlation analysis of the drought indices and hydrological variables in different seasons and climatic zones from 2000 to 2019. Our results suggested that areas with a significant increasing trend in precipitation (p < 0.05) occupied 13.4-20% of the study area from spring to autumn. The corresponding area for SM varied from 20 to 26%. A wetting trend was dominant in drylands despite TWS depletion. A drying trend appeared in southern North China Plain and southwest China, where severe droughts in 2009 and 2011 led to substantial declines in SM and TWS. SPI and SPEI worked better than scPDSI in characterizing meteorological drought at the seasonal timescale, given that the inter-annual variation of precipitation was in consistent with that of SPI and SPEI in approximately 70-80% of China. scPDSI was suitable for agricultural drought monitoring in different seasons, and agricultural drought was well characterized by SPI and SPEI only in the summer. SPI, SPEI and scPDSI both showed worse performances in quantifying drought trend and magnitude in drier regions and seasons characterized by a low aridity index. Thus, potential future expansion of global drylands calls for an integration of human impact into hydrological drought modeling.

Automated summary

In recent years, multiple drought indices have been developed to quantify drought characteristics and examine drought effects. Our study found that SPI and SPEI are better at characterizing meteorological drought at the seasonal timescale, while scPDSI is more suitable for agricultural drought monitoring. However, all three indices showed worse performances in quantifying drought trend and magnitude in drier regions and seasons characterized by a low aridity index.