Abstract: Pan evaporation (E_pan) is a critical measure of the atmospheric evaporation demand. Analyzing meteorological data from the Tropical Rainforest Comprehensive Meteorological Observation Field in the Xishuangbanna Tropical Botanical Garden (XTBG Meteorological Observing Station) based on physical models is helpful to improve our understanding of the state of the hydrological cycle in the Xishuangbanna tropical rainforest region. In this study, we investigated the long-term trend in E_pan using the observation data from 1959 to 2021. Moreover, correlation analyses of E_pan were performed, such as trend test, assessment of periodic properties and abrupt change analysis. Then, D20 E_pan data and related meteorological data from 1979 to 2008 were used to drive Penman‒Monteith and PenPan models for simulating E_pan. The partial derivative attribution method was used to analyze the dominant factors affecting E_pan. The results showed that E_pan exhibits obvious periodic changes, the 19a is the first primary period. In addition, there was a clear ‘evaporation paradox’ phenomenon in Xishuangbanna. E_pan showed a decreasing trend during both 1959-2008 and 2009-2018, and the decreasing trend reached a significant level with a rate of -3.404 mm·a^-2 during 1959-2008. Through comparative analysis, the PenPan model was considered more suitable for simulating E_pan in Xishuangbanna. In order to identify the main meteorological factors influencing E_pan, complete data from the D20 pan monitoring period, namely, 1979-2008, were selected for attribution calculations. The variations in the net radiation and saturated vapor pressure deficit are the main triggers that explain the ‘evaporation paradox’ phenomenon in Xishuangbanna.

Key words: pan evaporation, evaporation paradox, PenPan model, partial derivative method, driving factors