Detecting Model of Soil Total Nitrogen Content Based on NIR Technology and Wavelet Packet Analysis
会议名称：《亚洲精细农业与计算机农业应用联合大会》
会议日期：2009年
学科分类：082804[工学-农业电气化与自动化] 08[工学] 0828[工学-农业工程] 09[农学] 0903[农学-农业资源与环境] 090301[农学-土壤学]
关 键 词：Precision agriculture soil total nitrogen wavelet packet characteristic spectrum portable TN detector NIR
摘 要：It is the actual demand of agricultural production to develop the portable soil total nitrogen （TN） detector with higher accuracy. In order to meet the detecting accuracy demand of the portable TN detector based multi-band near-infrared reflectivity （NIR） technology, the NIR model detecting TN was developed based on wavelet packet analysis. In order to establish the TN detecting model with a finite number of near-infrared bands, 100 soil samples were collected for calibration and validation from the field. First, using the high-precision NIR detecting instrument to scan the target and obtaining the continuous spectra of soil samples in the laboratory. Secondly, using the method of wavelet packet analysis, the original signal of each soil sample was decomposed and reconstructed. Then the characteristic spectrum corresponding to each soil parameter was drawn form the original signal, and the partial least square （PLS） model for TN was established based on the drawn characteristic spectrum, which includes 24-band spectrum as the independent variables. Finally, due to the spectral feature of the characteristic spectrum of TN, nine near-infrared bands were selected and identified as the independent variables of the portable TN detecting instrument in accordance with the characteristics of each band’s contribution to the extent of the model. The TN content detecting model was established in the laboratory, and the determined R2 reached 0.700. It was showed that the accuracy of the TN detecting model was able to meet the needs of actual production. The research concluded that wavelet packet analysis could eliminate or substantially reduce the factors outside the parameters to the spectrum directly or indirectly, and the obstacles in establishing linear models for soil parameters were removed. It is feasible and potential to the real-time prediction of TN content.