基于导数光谱的枸杞叶片含水量遥感反演

Remote Sensing Retrieval of Leaf Water Content in Lycium barbarum L.Based on Derivative Spectra

以宁杞7号为试材,采用导数光谱分析法、吸收谷光谱参数提取法及光谱植被指数法,提取一阶导数光谱参数,构建一阶导数光谱指数,并参考前人研究成果,选取对植被含水量估测效果较好的光谱指数,将这些光谱参数或光谱指数与枸杞叶片含水量进行相关性分析,并建立回归模型对枸杞叶片含水量进行遥感反演。结果表明:前人构建的光谱指数GVWI、NDWI、MSI、R/ND和PRI虽可用于枸杞叶片水分含量遥感反演,但精度不高;一阶导数光谱参数D(925~975)、SD(1825~1925)、D(1275~1450)、SD(1275~1450)及D(1825~1925)也可用于枸杞叶片含水量遥感反演,但效果较差。相比较而言,一阶导数光谱指数中有15个指数均可用于枸杞叶片含水量遥感反演,反演效果较理想;其中,一阶导数光谱指数DSI(954-1387)为最优指数,基于该指数构建的模型y=566965x2-4719.5x+88.864为最佳模型,该模型决定系数、检验决定系数、均方根差和平均相对误差分别为0.6596、0.7604、0.6659、0.7164%。

One Lycium barbarum L.variety‘Ningqi No.7’was used as experimental material.First order derivative spectral parameters were extracted and first order derivative spectral indices were constructed using derivative spectral analysis,spectral parameters extraction of absorption valley and spectral vegetation index method,as well as the spectral indices with better estimation effect on vegetation water content were selected according to the previous research results.Besides,the correlation analysis of the above spectral parameters/indices and leaf water content of Lycium barbarum was carried out,then the regression model was calculated to retrieve the leaf water content of Lycium barbarum by remote sensing.It was found the spectral indices constructed by predecessors,GVWI,NDWI,MSI,R/ND and PRI could be used to retrieve the leaf water content of Lycium barbarum,but they had lower accuracy.The first order derivative spectral parameters D(925~975),SD(1825~1925),D(1275~1450),SD(1275~1450)and D(1825~1925)were not effective in retrieving Lycium barbarum leaf water content.In contrast,15 of the first order derivative spectral indices could be used for remote sensing retrieval of Lycium barbarum leaf water content,which had high accuracy and strong prediction ability.Among them,the first order derivative spectral index DSI(954~1387)was the optimal index on which the optimum model y=566965 x2-4719.5x+88.864 was constructed.The model’s determinant coefficient,test determinant coefficient,RMSE and ARE were 0.6596,0.7604,0.6659 and 0.7164%,respectively.The results will provide technical support for rapid and nondestructive monitoring of water content in the leaves of Lycium barbarum.