基于综合高光谱指数的区域土壤盐渍化建模——以平罗县为例

Regional soil salinization modeling based on comprehensive hyperspectral index: taking Pingluo County as an example

为建立综合高光谱土壤盐渍化模型,以干旱区典型的土壤盐渍化区域宁夏回族自治区平罗县为研究区,以植被和土壤的实测高光谱数据和土壤含盐量测量结果为基础数据,对高光谱数据进行平滑(S-G方法)和数学变换(倒数、对数和一阶微分等),并将其与土壤盐分数据进行相关分析,筛选对土壤含盐量响应最敏感的光谱波段,计算并选取最优高光谱植被指数和土壤盐分指数,采用多元线性回归方法建立高光谱土壤盐渍化模型.结果表明,以实测光谱反射率的对数一阶变换结果与土壤盐分指数和高光谱植被指数相结合作为自变量,土壤含盐量为因变量,所构建的多元线性回归模型效果最佳.该模型的预测值和实测值的相关性较好,R2=0.8279,通过99%的显著性检验;均方根误差为0.236 g/kg,相对分析误差为2.029,表明样本实测值与预测值之间的偏差较小,在一定程度上可以预测土壤含盐量.

In order to establish a comprehensive hyperspectral soil salinization model and,taking Pingluo County of Ningxia Hui Autonomous Region,a typical soil salinization area in the arid area,as the study area,combined with the hyperspectral data of vegetation and soil measured in the field and soil salinity data,hyperspectral data were smoothed(S-G)and transformed(countdown,logarithm and first order differential,etc.).The optimal bands corresponding to soil salt content were selected via correlating hyperspectral data with soil salinity data.The optimal hyperspectral vegetation index and soil salinity index were calculated and selected,and the hyperspectral soil salinization model established by the multiple linear regression method.The results showed that the logarithmic first-order transformation results of the measured spectral reflectance,combined with the soil salinity index and hyperspectral vegetation index as the independent variables;and with the soil salinity as the dependent variable,the prediction model established by multiple linear regression model was an optimal model.The correlation between the predicted value and the measured value of the model was good(R^2=0.8279),passing the significant level test of 0.01.The root mean squared error was 0.236 g/kg and relative percent deviation 2.029,indicating that the deviation between the measured and the predicted value was small,and the soil salt content could be predicted.