土壤总氮和总磷含量的高光谱遥感预测

用遥感技术监测土壤养分是精准施肥中的一个重要手段。为此,在对118个风干的土壤样本的室内反射光谱进行预处理的基础上,分析了土壤总氮和总磷浓度与预处理后的反射光谱的相关性。分别选用与总氮和总磷浓度具有较强相关性的3个波段(分别是684,724,1 890,2 002nm和2 342,1 092,639,2 262nm),用偏最小二乘方法建立了土壤总氮和总磷含量预测的高光谱分析模型。结果显示,利用预处理的反射光谱建立的土壤总氮与总磷模型的决定系数分别是0.842和0.622,其均方根误差分别是0.132g/kg和0.208g/kg。

Prediction for Total Nitrogen and Total Phosphorus Concentrations Using Hyper-spectral Remote Sensing

Hyper-spectral remote sensing could be a good tool for precision fertilization.The correlation on the soil total nitrogen and reflectance spectra preprocessed,the correlation on total phosphorus and reflectance spectra preprocessed were analyzed,using the laboratoty reflectance spectra of 118 soil samples.The prediction models of total nitrogen and total phosphorus based four wave bands(four wave bands for the prediction model of total nitrogen were 684,724,1 890,2 002nm four wave bands for the prediction model of total phosphorus were 2 342,1 092,639,2 262nm) were established,using partial least square(PLS) method.The results show that the determination coefficients R2 of total nitrogen and total phosphorus were 0.842 and 0.622,respectively,the RMSE of total nitrogen and total phosphorus were 0.132g/kg and 0.208g/kg,respectively.