基于小波变换的光谱分析法预测土壤总氮含量

Predict Soil Total Nitrogen Content with Near-infrared Spectroscopy Treated by Wavelet Transform

为了快速检测土壤总氮含量,本文采用小波(wavelet)变换方法,分析了山西省关帝山土壤中总氮(N)的近红外光谱信息.结果表明:对总氮敏感的波段为835 nm^845 nm,1 400 nm^1 570 nm,1 775 nm^1 950 nm,2 000 nm^2 160 nm;总氮NIRS法与实验室标准法测定值之间的决定系数R2为0.981 9;说明通过小波变换滤波,选择敏感波段,用偏最小二乘回归方法预测土壤总氮含量是可行的.

In this paper,we analysed the near-infrared spectroscopy signals of soil total nitrogen content in Shanxi Guandi mountain based on wavelet transform.The results showed that sensitive band of the total nitrogen for the 835 nm- 845 nm,1 400 nm- 1 570 nm,1 775 nm- 1 950 nm,2 000 nm- 2 160 nm.The correlation coefficient R2 was 0.9819 between total nitrogen NIRS method and the standard laboratory method.Therefore,it is feasible to predict soil total nitrogen content through wavelet filtering,sensitive band selection using partial least-squares regression method.