As one of the extreme climatic events, the frequency and intensity of drought have great impacts on regional water resource. Water is a main limiting factor for plant growth in arid and semi-arid regions. Therefore, it is of great scientific significance to explore the spatiotemporal variations and future tendency of drought for the ecological environment in the Loess Plateau. Based on grid data of monthly precipitation and temperature from 1986 to 2019, we calculated standardized precipitation evapotranspiration index (SPEI) and drought frequency. The spatiotemporal patterns and its variations were analyzed at the seasonal and annual scales in the Loess Plateau using the Mann-Kendall test and Sen's slope estimation method. Finally, the future trend of drought was analyzed in the Loess Plateau by the NAR neural network combined with Hurst index. Results showed that the trend of aridification became more significant in the Loess Plateau, and that the frequency of droughts events exhibited great spatial variations at the interannual and seasonal scales during the study period. Specifically, the highest frequency of drought in the interannual, spring and winter was found in the southeast and west of the Loess Plateau, whereas the frequency of drought in summer and autumn was higher in the northwest. The frequency of moderate drought was the highest in summer compared with other seasons while the frequency of slight drought was the highest in interannual and other seasons. The Loess Plateau showed a trend of aridification in spring and summer, but this trend in autumn and winter became weaker in most areas of the study area. The SPEI value in the interannual, spring, and summer exhibited a decline trend in a future period in the Loess Plateau. The aridification would be enhanced. The Hurst index value was the largest and the persis-tence of its change remained stronger in summer. The possibility of continuous drought in summer would be higher than that in other seasons in the future.
干旱作为极端气候事件之一,其频率和强度的变化影响到区域水资源,而干旱半干旱区植物生长的主要限制因素是水分,因此,研究黄土高原干旱时空特征及未来变化趋势对当地的生态环境具有重要意义。本研究基于1986—2019年降水和温度逐月格点数据,计算标准化降水蒸散发指数(SPEI)和干旱发生频率,并运用Mann-Kendall检验和Sen斜率估计方法,探讨了黄土高原的年、季尺度干旱时空分布及变化特征,最后利用NAR神经网络结合Hurst指数对黄土高原未来干旱趋势进行预测。结果表明: 研究期间,黄土高原总体呈现干旱化趋势,且年际和季节尺度的干旱发生频率在空间上差异较大。其中,年际、春季和冬季以黄土高原东南部和西部干旱发生频率最高,夏季和秋季以西北部干旱发生频率最高。夏季以中度干旱发生频率最高,年际及其他季节以轻度干旱发生频率最高。黄土高原春、夏季呈现干旱化趋势,秋、冬季研究区大部分区域干旱趋势减轻。黄土高原年际、春季、夏季的SPEI值在未来一段时间内仍处于下降趋势,即干旱化趋势加重,且夏季的Hurst指数最大,持续性变化最强,未来持续干旱的可能性高于其他季节。.
Keywords: Hurst index; Loess Plateau; drought; neural network; standardized precipitation evapotranspiration index.