Research on Soil Nutrient Content Estimated Model by Hyperspectral Remote Sensing Data

【Objective】 To obtain the real-time, non-contact and non-destructive information of main nutrients (TN, TP and TK) of farmland soil quickly.【Method】Soil nutrient content model was established by spectral feature analysis techniques and sensitive wavebands.【Result】NDI-based estimates of the soil nutrient content prediction model with exponential function model(Y^TN =0.000,5 e^{4.700,3 xNDI})was the best to estimate TN; a cubic function model(Y_TP =802.27 x_NDI³-412.32 x_NDI²+72.357 x_NDI-3.318,9)was the best to estimate TP; a cubic function model(Y_TK =80,189 x_NDI³-11,471 x_NDI²+490.57 x_NDI+13.879)was the best to estimate TK content.【Conclusion】The results showed that the hyperspectral remote sensing quantitative model based on normalized spectral index (NDI) can be used to invert the content of TN, TP and TK in soil and achieve a good prediction effect through the precision evaluation of the model and repeated verification in fields.