Identifying multivariate controls of pan evaporation variations in subregions of Yangtze River Basin using multiple wavelet coherence

Abstract

AbstractIn this study, temporal variations of pan evaporation (PE) and its optimal combination of controlling factors are investigated in the subregions of Yangtze River Basin (YRB) by using multiple wavelet coherence (MWC). For this purpose, the YRB is firstly divided into nine relatively homogeneous PE subregions by combining empirical orthogonal functions with fuzzy C‐means clustering. After regionalization, temporal variations of regional PE are analysed in each of subregions. Results reveal that regional annual PE has an upward trend in all subregions, and seasonal PE in almost the subregions for four different seasons. The wavelet transform coherence (WTC) analysis shows that vapour pressure deficit (VPD) is the primary controlling factor of PE variations in all subregions. In single factor analysis, VPD, sunshine duration (SSD) and 0 cm ground temperature (ZGT) have a higher covariance with PE than other influencing factors. For two‐factor analysis, the APS (air pressure)‐VPD combination, and for three‐factor analysis, the AWS (average wind speed)‐ ZGT‐VPD combination are the most common best explanatory variables on PE variations. A comparison of the results from the WTC and MWC indicates that PE variations can be explained more accurately by one, two, or three controlling factors. The combination of factors that optimally explains PE variations largely varies with subregions but with VPD as a dominant factor either alone or in combination for all subregions. Moreover, the optimal factor combination for a subregion is not necessarily combined by the first, second and/or third dominant factors for that subregion. Our results can provide useful insights into how multiple temporal dependency of PE is linked with meteorological factors in subregions of the YRB, particularly for PE modelling under changing climate conditions.