【摘要】 为评估冬小麦覆盖地不同生育期及深度的盐分反演模型适用性，基于无人机多光谱遥感数据提取32种光谱指数，并同步采集拔节期、抽穗期和灌浆期0～20、20～40和40～60 cm深度的土壤含盐量。采用递归特征消除-极端梯度提升算法（RFE-XGBoost）筛选关键光谱变量，结合随机森林（RF）构建盐分反演模型。结果表明，拔节期为最佳反演生育期，0～20 cm为最佳反演深度。RF模型预测性能最优，其决定系数（R~2）、均方根误差（RMSE）和平均绝对误差（MAE）分别为0.748、0.029%和0.024%。本研究为冬小麦多生育期及多深度土壤含盐量估算提供了科学依据，对精准农业管理和土壤盐渍化监测具有重要意义。更多还原

【Abstract】 To evaluate the applicability of salt content inversion models for winter wheat at different growth stages and depths under mulching conditions,32 spectral indices were extracted based on unmanned aerial vehicle(UAV) multispectral remote sensing data,and soil salinity at depths of 0～20 cm,20～40 cm,and 40～60 cm was simultaneously collected during the jointing stage,heading stage,and filling stage.The recursive feature eliminationextreme gradient boosting algorithm (RFE-XGBoost) was used to select key spectral variables,and the random forest (RF) model was constructed for salt content inversion.The results showed that the jointing stage was the best growth stage for inversion,and the 0-20 cm depth was the best inversion depth.The RF model had the best prediction performance with a coefficient of determination (R~2) of 0.748,a root mean square error (RMSE) of 0.029%,and a mean absolute error (MAE) of 0.024%.This study provides a scientific basis for estimating soil salinity at multiple growth stages and depths of winter wheat and is of great significance for precision agriculture management and soil salinization monitoring.更多还原