Trend Analysis of Seasonal Precipitation (1960-2013) in Subregions of Hunan Province, Central South China Using Discrete Wavelet Transforms

摘要

In this study, trends of seasonal precipitation are investigated in subregions of Hunan Province in China over 1960-2013 by discrete wavelet transform (DWT) and Mann-Kendall (MK) tests. The main purpose of trend analysis is to detect the most dominant periodic components affecting the observed trends of seasonal precipitation in each subregion. For this purpose, the homogeneous precipitation regions are first delineated by combining rotated empirical orthogonal functions with agglomerative hierarchical clustering. Results suggest dividing the precipitation of Hunan Province into five subregions for winter and three subregions for spring, summer, and autumn. Delineated subregions for each season show strongly coherent spatial patterns, all of which are along the strip extending in the southwest-northeast direction. After regionalization, areal seasonal precipitation series in each subregion are decomposed into several detail and approximation component subseries, and then the decomposed subseries are analyzed by three types of MK tests. Results reveal that winter and summer precipitations experience an increasing trend in all of the divisions-winter precipitation in southeast-central and southern Hunan and summer precipitation in southwest-central Hunan especially exhibit a statistically significant tendency. However, spring and autumn precipitations show a nonsignificant decreasing trend in all of the divisions. Results from DWT and sequential MK analyses suggest that detail component 1 (plus approximation) of seasonal precipitations in all subregions is the most dominant periodic mode affecting their observed trends, except for summer precipitation in the subregion of southern Hunan, where detail component 2 is the most dominant one. Therefore, short-term periodicity (2-4 yr) is the most influential in affecting the observed trends of seasonal precipitation over Hunan Province.