基于西藏地区38个气象站点1981—2019年逐日气象资料,采用Penman-Monteith模型、趋势分析、Morlet小波分析、Mann-Kendall检验、经验正交函数法（Empirical orthogonal function,EOF）和灰色模型探究潜在蒸散量（ET₀）的时空演变规律以及未来ET₀的变化趋势。结果表明：在时间尺度上,西藏地区ET₀平均为597.12 mm,1981—2001年呈显著的减少趋势、2002—2019年呈显著的增加趋势（P<0.01）;西藏全区及五大气候区年均ET₀均呈现增加趋势（除东南部）,且均以33 a的周期震荡为第一主周期。在空间尺度上,年ET₀主要呈现由中部中心向西南和东南逐步递减的分布特征,高值中心集中在沿江一线,低值中心位于南部地区。年ET₀发生突变的站点主要分布在南部边缘地区、沿江一线和东北部,发生时间集中在20世纪80年代。构建的GM（1, 1）预测模型预测精度均值为87.85%,可用于西藏年均ET₀日期序列的中长期预测,预测结果显示,除东南部年ET₀的预测值有明显下降,其他区域均呈现上升态势。

This paper aims to provide a scientific basis for water-saving irrigation, the rational allocation of water resources, and drought monitoring in Tibet, China. We use the Penman-Monteith model, trend analysis, Morlet wavelet analysis, Mann-Kendall tests, the empirical orthogonal function, and gray models to explore the spatiotemporal evolution of potential evapotranspiration (ET₀) in Tibet and the central and western regions of Nagqu as well as the trend of future ET₀ changes based on daily meteorological data collected from 1981 to 2019 at 38 meteorological stations in the region. The results demonstrate that on the time scale, the average ET₀ in Tibet is 597.12 mm. It also exhibited a significant decreasing trend from 1981 to 2001 and a significant increasing trend from 2002 to 2019 (P<0.01). In these two distinct time periods, high evapotranspiration was concentrated in the central and western areas of Nagqu and along the river line. In the first mode (EOF1), the annual ET₀ change trend in Tibet had a high degree of consistency in space, mainly exhibiting a gradually decreasing distribution from the center of the central area to the southwest and southeast. The high-value center is concentrated along the river line, and the low-value center was located in the southern region. Concerning space, the positive center of central and western Tibet and the negative center of southeastern Tibet are reversed. The stations with positive Kendall trend coefficients are mainly concentrated in the central and western regions of Nagqu and the southern marginal area, where the climate has been warming and drying. The sites exhibiting abrupt changes in annual ET₀ are mainly distributed in the southern marginal area along the river line and the northeast and were concentrated in time. In the 1980s, two mutations occurred in the potential evapotranspiration at Nagqu and Dangxiong stations, whereas one or no mutations occurred at each of the remaining stations. The annual average ET₀ both of Tibet as a whole and of the five major climate zones exhibited an increasing trend (except in the southeast), and the 33 year cyclic oscillation was the first main cycle. Additionally, there was a cycle of 11 a in the whole region and along the river line. The midwest of Nagqu had cycles of 22 a and 3 a, the southern marginal area had cycles of 8 a and 12 a, the northeast had a cycle of 8 a, and the southeast had main cycles of 22 a and 12 a. The constructed gray model (GM; 1, 1) used for prediction had gray parameters a≥-0.3 and a∈(-2, 2), and the level ratio was in a smooth state, meaning it can be used to predict future ET₀. The average prediction accuracy was 87.85%, so we predicted the medium- and long-term average annual ET₀ date series for Tibet. The forecast results revealed that except for the significant decline in the forecast value of the annual ET₀ in the southeast, all other regions are exhibiting an upward trend, and the southern edge shows the steepest increase.