Research On Spectrum Estimation Of Soil Nutrient Content Under Different Texture

In recent years, blindly fertilization lead to environmental problems are increasingly serious, the purpose of precise fertilization is to identify and quantify the differences in the field, understand the possible impact of these differences and the causes of differences, and carries on the variable management according to these differences, finally obtain good economic and ecological benefits. This study by using spectrum analysis technology is rapid, simple, nondestructive and the characteristics of green, different texture of soil total nitrogen content in Xinjiang by spectral prediction model is built, and the accuracy of the model is validated by actual samples.Using field effect of practical application, a better foundation for crops and precision of variable rate fertilization. Main conclusions are as follows:1. The different soil texture contrast test results show that the major differences between the spectral reflectance of the soil texture. The fine particles of soil(sand), spectral curve is smooth, with high reflectivity; The coarser soil, gravel soil spectral curve and volatile, especially in the 1300 nm, 1800 nm and 2200 nm has obvious absorption valley. Different nitrogen, phosphorus and potassium content in soil and its good negative correlation exists between the spectral reflectance;2. Soil spectral reflectance and soil total nitrogen, total phosphorus, total potassium content is a good correlation between. Soil total nitrogen content and the original spectral reflectance at 1867 nm is significantly negative correlation, the correlation coefficient reaches 0.9392; To reflectivity inverse transform the second derivative of the best correlation with TN and spectral reflectance from bottom of first order differential and second order differential further strengthened and the correlation of soil total phosphorus content, the best correlation with TN and content of soil total potassium(TK) and the relation with first derivative spectral reflectance transformation after the optimal, at 1517 nm 0.9979 very significant positive correlation.3. To estimate the soil total nitrogen content of exponential function model of multiple correlation coefficient(R2 = 0.7982), the highest in exponential function relation model(YTN =0.0005e4.7003xNDI); the optimal model for predicting total nitrogen. Total phosphorus content of soil phosphorus estimating model, three times a yuan function model after the highest correlation coefficient(R2 = 0.5631); Three times a yuan function(YTP =802.27xNDI3-412.32xNDI2+72.357xNDI-3.3189), the optimal; Soil total potassium content estimation model, three times a yuan function model of the multiple correlation coefficient(R2 = 0.5150) is highest, three times a yuan function relation model of the prediction effect is the best(YTK=80189xNDI3-11471xNDI2+490.57xNDI+13.879).4. The estimation of soil total nitrogen content of model prediction effect is best, prediction accuracy is 78.1430%, and the relative error minimum, 21.8570%; The second is the estimate of quantity of soil total phosphorus model prediction accuracy is higher, up to 75.6690%; Among them, the soil total potassium content between the predicted and the measured values of the relative error is larger, at 29.7039%, the estimation of the model prediction accuracy is low, at just 70.2961%. Spectroscopy is used to analyse the soil characteristic information of real-time monitoring, rapid extraction, accurate prediction and inverse modeling is feasible.