基于随机森林的极端气候事件对黄淮海地区小麦产量的影响

Effects of Extreme Climate Events Based on Random Forest on Wheat Yield in Huang-Huai-Hai Region

摘要: 本文以黄淮海地区为主要研究区，基于2001~2018年统计年鉴产量数据和生长季NPP数据，选取了14个表征极端气候的指标，运用随机森林模型和Spearman相关分析法，评估了各气候指标对产量的重要性影响程度，并依据评估结果建立了2~5月各月份主要影响指标与生长季NPP之间的关系。研究得出冬小麦产量对研究区各月份最重要影响指标的响应机制表现为：2~4月份高温均有利于研究区部分地区的冬小麦生长；而5月份的高温则不利于研究区中南部部分地区冬小麦的生长。2月份的强降水不利于研究区中部部分地区冬小麦的生长；而3~5月份降水量和降水强度的增加均有利于研究区中南部部分地区旱情的缓解。4月份干旱整体不利于研究区中南部部分地区冬小麦的生长；而2、3、5月份干旱的发生对产量的影响在空间上存在差异：其中2、3月份适度干旱均有利于河北部分地区冬小麦的生长，不利于河南西南部冬小麦的生长；5月份，干旱在一定程度上有利于研究区中部部分地区冬小麦的生长，推测是以上地区5月份的降水强度呈上升趋势，整体趋于湿润，适当的干旱反而保证了小麦含水率下降、干重增长，而研究区南部地区则显示干旱不利于冬小麦的生产。

Abstract: Taking the Huang-Huai-Hai region as the main research area, based on the yield data in the statistical yearbook and the NPP data in the growing season from 2001 to 2018, this study selects 14 indicators representing extreme climate, evaluates the importance and impact of various climate indicators on yield by using random forest model and Spearman correlation analysis method. Based on the evaluation results, the relationship between the main impact indicators and NPP in the growing season from February to May was established. The results show that the response mechanism of winter wheat yield to the most important impact index of each month in the study area is as follows: the high temperature from February to April is conducive to the growth of Winter Wheat in some areas of the study area; the high temperature in May is not conducive to the growth of Winter Wheat in some parts of the central and southern part of the study area. The heavy rainfall in February is not conducive to the growth of Winter Wheat in some parts of the central part of the study area; the increase of precipitation and precipitation intensity from March to May is conducive to the mitigation of drought in some parts of the central and southern part of the study area. The drought in April was not conducive to the growth of Winter Wheat in some parts of the central and southern part of the study area; there are spatial differences in the effects of drought on yield in February, March and May: moderate drought in February and March is conducive to the growth of Winter Wheat in some areas of Hebei, but not in the southwest of Henan; in May, drought is conducive to the growth of Winter Wheat in some parts of the central part of the study area to a certain extent. It is speculated that the precipitation intensity in the above areas in May shows an upward trend and tends to be humid as a whole. Appropriate drought ensures the decline of wheat moisture content and the growth of dry weight, while the southern part of the study area shows that drought is not conducive to the production of winter wheat.