基于机载多光谱的冬小麦返青期土壤墒情反演

准确快速得获取冬小麦地块的土壤墒情,可为高效利用水资源、实现精准灌溉提供参考。为此,特在江苏省张家港市获取返青期冬小麦种植区的无人机多光谱遥感数据,并同步测定2个深度（10 cm和20 cm）的土壤墒情,通过遥感图像提取光谱反射率,计算归一化植被指数（NDVI）、增强型植被指数（EVI）和垂直干旱指数（PDI）,进行共线性分析后,分别运用逐步回归法、岭回归法和偏最小二乘法,构建针对不同深度土壤墒情的反演模型,并基于最佳反演模型绘制试验区不同深度土壤的墒情反演图。结果表明,用逐步回归法构建的模型在10、20 cm深度土壤墒情反演中的决定系数分别达到了0.885、0.782,建模精度最优,且针对10 cm深度土壤墒情的反演效果优于20 cm。研究结果可为冬小麦返青期土壤墒情监测方法的选择提供参考。

Inversion of soil moisture content of winter wheat at turning green period based on multispectral remote sensing by unmanned aerial vehicle

Accurate and quick determination of soil moisture of winter wheat plot could provide references for efficient water utilization and precision irrigation. In this study, the multispectral remote sensing data of winter wheat at turning green period were obtained by unmanned aerial vehicle in Zhangjiagang City, Jiangsu Province. The soil moisture at two depths (10 cm and 20 cm) was measured simultaneously. Spectral reflectance was extracted from remote sensing images, and normalized difference vegetation index, enhanced vegetation index, perpendicular drought index were calculated. After collinearity analysis, stepwise regression, ridge regression and partial least squares regression methods were used to construct soil moisture content inversion models at different soil depths, then the optimal inversion model was adopted to draw soil moisture inversion maps. The results showed that the determination coefficient of models constructed by the stepwise regression method for 10, 20 cm soil depths were 0.885 and 0.782, respectively, of which the inversion precision was the highest. The inversion performance for 10 cm soil depth of all the constructed models were better than that of the 20 cm soil depth. The present study could provide references for the selection of soil moisture monitoring of winter wheat at turning green period.