GF-5高光谱遥感影像的土壤有机质含量反演估算研究

本文基于高光谱遥感影像对土壤有机质含量进行反演估算，以哈尔滨与兴安盟交界处的平原地表土壤为试验对象.首先，基于辐射和几何校正等预处理的高分五号（GF-5）高光谱遥感影像，依据五点采样法采集影像覆盖范围的地表土壤样本共100组，在实验室内进行理化分析等一系列处理获取样本土壤有机质含量数据.然后，运用偏最小二乘法建立高光谱影像土壤沙化指数、土壤退化指数、归一化亮度指数和土壤盐分指数反演土壤有机质含量的估算模型.比较基于原始反射率数据、一节微分反射率数据和4种土壤指数构建的反演模型的预测精度，通过65%的建模样本和35%的预测样本验证表明，反演模型中基于土壤指数建立的反演模型的预测验证精度最高，预测集验证中ρ为0.816，RMSE为1.7287.并将该反演模型运用到高光谱影像的土壤有机质含量的反演估算，实际测量的SOM与影像反演SOM含量变化趋势一致，相关性达到80.023%以上，验证了模型的反演估算精度.

Inversion estimation of soil organic matter content based on GF-5 hyperspectral remote sensing image

Based on the hyperspectral remote sensing images inversion, this paper estimated the soil organic matter content of the plain in the Harbin and Xing’an League boundary. Firstly, from the Gao-Fen 5(GF-5) hyperspectral remote sensing image, which has been pre-processing such as radiation and geometric correction, 100 groups of surface soil samples were collected by the five-point sampling method, and a series of physical and chemical analysis in the laboratory were performed to obtain the soil organic matter content. Then, the partial least squares regression method was applied to establish an estimation model of soil organic matter content by hyperspectral image soil desertification index, soil degradation index, normalized brightness index and soil salinity index. The prediction accuracy of the inversion model based on the original reflectance data, one differential reflectance data, and four soil indices were compared. From 65% modeling samples and 35% prediction samples it showed that the inversion model based on the soil index in the inversion model had the highest prediction accuracy. In the verification of the prediction group, ρ was 0.816 and the RMSE was 1.7287. Finally, the inversion model was applied to the inversion estimation of soil organic matter content by hyperspectral imagery. The actual measured SOM was consistent with the image inversion SOM content change trend, and the correlation reached 80.023%, which verified the accuracy of the model's inversion estimation.