黄河源区近地表冻结状态变化及其对径流的影响OA
Changes in near-surface freeze-thaw state and their impact on stream flow in the source region of the Yellow River
基于1982-2020年黄河源区0～40 cm土壤温度数据,分析黄河源区近地表土壤温度和土壤冻结状态变化特征,结合气象与植被数据集,采用随机森林和SHAP(Shapley Additive Explanations)方法解析近地表土壤冻结状态变化对径流的影响.结果表明:黄河源区年径流变化在2002年发生显著转变,由前期显著下降转变为后期显著增加.同期,0～40 cm土壤升温模式在垂直方向、季节尺度和空间格局上均发生一定转变,表现为土壤分层冻结面积显著缩减,冬季冻结深度明显变浅.基于随机森林和SHAP分析发现,径流对土壤分层冻结状态变化的响应具有明显的非线性特征.在表征土壤分层冻结状态的因子中,1982-2002年径流主要受到0 cm和10 cm深度冻结面积影响,相对影响占比分别为8.3％和7.2％,而2002-2020年起主要作用的因子为0 cm和40 cm深度的冻结面积,相对影响占比分别为10.6％和8.4％.此外,径流年际波动对0～10 cm深度土壤冻结状态的年际波动较为敏感.研究揭示了气候变暖背景下黄河源区近地表土壤温度变化和温度变化引起的冻结状态变化及其对径流的影响,为寒区水文过程研究提供了科学依据.
Based on soil temperature data at depths of 0-40 cm,we investigate the characteristics of soil tempera-ture and freeze-thaw state dynamics in the Yellow River source region and their impacts on streamflow from 1982 to 2020.Using 1-km resolution gridded soil temperature data(0-40 cm depths)interpolated from station ob-servations using a thin plate spline method,we analyzed long-term trends in soil temperature and associated changes in soil freezing states.In addition,we integrated meteorological variables(e.g.,precipitation,air tempera-ture,and evapotranspiration)and normalized difference vegetation index(NDVI)datasets to quantify the impact of soil freezing state variation on streamflow using random forest and Shapley Additive Explanations(SHAP).The results show a marked shift in the streamflow trend in the Yellow River source region:A sharp decline oc-curred during the earlier period(1982-2002),with an average annual reduction of 6.11×108 m3,followed by a no-table increase during the later period(2002-2020),at an average rate of 6.02×108 m3 per year.Soil warming at 0-40 cm depths was accompanied by a substantial decrease in the frozen soil area across all layers and notable freez-ing depth shallowing in winter.Random forest modeling combined with SHAP interpretation revealed a nonlinear relationship between stratified soil freezing states and streamflow.Among the variables representing soil freezing states,the streamflow during 1982-2002 was mainly influenced by the frozen areas at 0 cm and 10 cm depths,with relative contributions of 8.3％and 7.2％,respectively.Meanwhile,during 2002-2020,the primary influenc-ing factors shifted to the frozen areas at 0 and 40 cm,with relative contributions of 10.6％and 8.4％,respectively.Further,the interannual variations in streamflow were more sensitive to fluctuations in the surface soil freezing state(0-10 cm)than in the deeper freezing state(15-40 cm).This study elucidates the changes in near-surface soil temperature and associated freeze-thaw dynamics under climate warming and their effects on streamflow,pro-viding key insights for hydrological process studies in cold regions.
鲍玥颖;朱求安;罗栋梁;侯玉宇;于冬雪;方秀琴;任立良
河海大学水文水资源学院,江苏 南京 210024河海大学地理与遥感学院,江苏 南京 211100中国科学院西北生态环境资源研究院,冰冻圈科学与冻土工程全国重点实验室,甘肃 兰州 730000河海大学地理与遥感学院,江苏 南京 211100河海大学水文水资源学院,江苏 南京 210024河海大学地理与遥感学院,江苏 南京 211100河海大学水文水资源学院,江苏 南京 210024
土壤温度随机森林SHAP黄河源区
soil temperaturerandom forestSHAPsource region of the Yellow River
《干旱区研究》 2026 (3)
527-538,12
国家重点研发计划项目(2024YFF0808704)国家自然科学基金项目(42041005,U2243203)
评论