Evolution of Wet and Dry Spells Based on Original and Corrected Precipitation Data in Southwest China, 1961–2019

Abstract

Gauge-measured precipitation data have long been recognized to underestimate actual precipitation due to wind-induced error, trace precipitation, and wetting loss, which affects the spatial and temporal characteristics of precipitation. In this study, we examined spatial and temporal differences in wet and dry spell indices based on original (Po) and corrected (Pc) precipitation data and their correlations with large-scale circulation indices (LSCIs) in Southwest China during 1961–2019. The main conclusions were: (1) Pc-based trends in wet/dry spell indices were generally more pronounced than Po-based. Specifically, when Pc-based, more stations had significant changes in the MWS, MLWS, MPWS, PWS95, FWW, FDW, MDS, MLDS, NLDS, and DDS95 indices, while fewer had significant changes in the NWS, NDS, FDD, and FWD indices. (2) Spearman’s results showed that more LSCIs were significantly related to the Pc-based wet/dry spell indices than Po-based. Po-based and Pc-based MWS, Po-based MDS, and Pc-based NLDS were significantly related to the most LSCIs. Therefore, taking them as examples, wavelet transform coherence (WTC) and partial wavelet coherence (PWC) were used to explore the coherence with LSCIs. WTC results showed South Asian Summer Monsoon Index (SASMI) + Po-based MWS, Arctic Oscillation (AO) + Po-based MDS, SASMI + Pc-based MWS, Asia Polar Vortex Intensity Index (APVI) + Pc-based NLDS exhibited the most obvious periodic resonance with main resonance periods of 2.13~7.8 year, 2.19~10.41 year, 2.13~12.13 year, 2.75~18.56 year, respectively. Since WTC may arbitrarily ignore the interaction between LSCIs, PWC is adopted for further analysis. PWC results showed the coherence of AO +Po-based MDS significantly increased after eliminating the Nino Eastern Pacific index (NEP) influence, with the main resonance period of 6.56~18.56 year. This study clearly demonstrated that corrected precipitation data should be used to improve the accuracy of drought assessments, climate models, eco-hydrological models, etc.