基于高光谱数据的农田土壤养分信息提取研究

By applying high spectral resolution remote sensing technology,this paper focuses on the monitoring of agricultural soil fertility status with hyperspectral remote sensing,obtaining the hyperspectral reflectance information of soil fast, effectively without contact and destruction with the application of ASD Field Spec Pro VNIR 2500 Spectrum radiometer. This paper also reveals the physical components of soil and the internal relation of its spectrum causes by using spectral analysis technology and various forms of soil reflectance mathematical transformation to identified the most representative sensitive waveband. Regression analysis is done about such items as organic matter content, moisture, total nitrogen, total phosphorus, total potassium of soil and its spectral reflectance characteristics parameters statistically in order to build a relevant mathematical model that predicts soil nutrient status which is based on normalized spectral index NDI. This paper also conducts accuracy test and field test on the quantitative model through the correlation analysis between the measured and predicted values. Therefore,the best estimation model which indicating the soil characteristic parameters is selected, followed by Interpolation Inversion Mapping. The following is about the study of space distribution and hyperspectral remote sensing quantitative monitoring which helps us acquire information with a hight speed on soil organic matter, moisture content, total nitrogen, total phosphorus and total potassium content.Studies shows that the application of hyperspectral remote sensing technology in terms of analysis, simulation, evaluation and prediction of soil characteristic parameters promotes the usage of hyperspectral remote sensing technology in the process of remote sensing diagnosis of farmland soil , offering technical support in space distribution monitoring and the management of water and fertilization. Meanwhile the course of quantitative evaluation research is promoted and the application of remote sensing technology in scientific crop plantation and crop condition monitoring is realized. This,therefore, provides precision agriculture in Xinjiang with scientific evidence.

本研究主要以农田土壤养分信息状况探测技术与分析处理为主要研究内容，拟利用高光谱遥感技术探索新疆绿洲农区应如何快速、有效、非接触、非破坏性地获取田间土壤特征信息。通过采用光谱分析技术，运用各种土壤反射率数学变换形式，找出最具代表性的敏感波段，揭示土壤养分特征信息与其光谱成因机理之间的内在联系；建立土壤信息的定量预测模型，通过实测值与预测值的相关分析进行精度检验与田间验证，筛选出最佳估算模型，并进行插值反演填图和农田养分分区管理应用的尝试。此研究成果将促进遥感技术在农田土壤肥力诊断中的应用，为土壤养分信息空间分布状况监测与分区管理提供技术支持，推进遥感技术在作物科学种植与田间农情监测中的应用，为新疆实施精准农业提供科学理论参考。

本研究主要以农田土壤养分信息状况探测技术与分析处理为主要研究内容，利用高光谱遥感技术探索新疆绿洲农区应如何快速、有效、非接触、非破坏性地获取田间土壤特征信息。通过采用光谱分析技术，运用各种土壤反射率数学变换形式，找出最具代表性的敏感波段，揭示土壤养分特征信息与其光谱成因机理之间的内在联系；建立土壤信息的定量预测模型，通过实测值与预测值的相关分析进行精度检验与田间验证，筛选出最佳估算模型，并进行插值反演填图和农田养分分区管理应用的尝试。此研究成果将促进遥感技术在农田土壤肥力诊断中的应用，为土壤养分信息空间分布状况监测与分区管理提供技术支持，推进遥感技术在作物科学种植与田间农情监测中的应用，为新疆实施精准农业提供科学理论参考。