Hydroclimatic representativeness and reassessment of the predictive skill to reconstruct precipitation by the Dryness/Wetness Index from Chinese historical documents

Abstract

Study region

China

Study focus

The Dryness/Wetness Index (DWI) data from China is widely utilized in palaeohydroclimate research. The initial part of this data, covering the years 1470–1979 (original DWI), is primarily generated from records of droughts/floods in historical documents. However, the extended data over 1980–2000 (precipitation-based DWI) is derived entirely from instrumental precipitation measurements. To date, there is no research on the regional differences in hydroclimatic representativeness of the original DWI. Moreover, when reconstructing precipitation using the combined 1470–2000 DWI data as a proxy and calibrating it with instrumental precipitation data post-1950, the overestimated reconstruction skill has not been evaluated. Therefore, utilizing data on crop yield reductions due to drought/flood disasters from 1978 to 2008, we establish the disaster-based DWI following the same method as the original DWI and explore its representativeness and reconstruction skill.

New hydrological insights for the region

Disaster-based DWI is more sensitive to precipitation in East Central China, with the seasonal window primarily lasting for five months and distributed between April and September. In contrast, it reflects soil water in the Northeastern, Southeast coastal, and Western regions in China. Additionally, by comparing the average predicted R-Squared of summer precipitation reconstruction using precipitation-based DWI and disaster-based DWI (58.6 % and 45.0 %, respectively), we identify an average overestimation of 13.6 %. Even after excluding this inflated R-Squared, the disaster-based DWI remains a highly reliable proxy for precipitation.