土地资源的高效管理对土地资源与生态环境监测提出更高的要求,高光谱遥感图像精细的光谱线可提高土地动态监测的精度、广度和深度,在土地资源与土地质量监测中具有广阔的应用前景。本文基于高光谱遥感技术,详细探讨了土地质量指标,主要是土壤有机质土壤水分信息提取的技术流程和关键技术。通过地物光谱分析认为,有机质含量越高的土壤,其光谱发射率越低;重度沙化土地实验室光谱曲线光谱反射率最高,依次排序为中度沙化、轻度沙化和无退化土地;重度水蚀土地光谱反射率较高,但中度水蚀土地与轻度水蚀和无退化土地实验室光谱曲线无明显差异。在光谱分析基础上,利用多元统计分析技术建立了土壤光谱在677、1202、2074和1509nm四个波段的有机质反演模型和在1423、1524、1746nm三个波段的预测土壤含水量的回归模型;并将有机质反演模型应用于宜兴成像光谱遥感数据中,进行宜兴试验区土壤有机质参数成图。最后探讨了反演模型应用推广所存在的问题。

With the development of land management from emphasizing quantity management to quality and ecology protection, higher demand for monitoring land resource and environment was arisen. The hyperspectral remote sensing image which has accurate and detail spectrum provides detailed surface information and improves precision, extent and the depth of dynamic monitoring on land use and will be widely applied in monitoring on land quality. This paper puts emphasis upon making a study of information extraction technology and methods for monitoring land quality indicators, including soil organic matter and soil moisture. Based on analyzing the spectrum of soil in Yixing and Hengshan field area, we think that the intensity of spectrum reflection is low when the concentration of soil organic matter is high. The highly desertified soil has the highest intensity of spectrum reflection measured in the laboratory, followed by moderately, slightly, and non-desertied soil. The highly water-eroded soil has relatively high intensity of spectrum reflection, but the difference in the measured intensity of spectrum refection of moderately, slightly and non water-eroded soil in the laboratory was not significant. According to above analysis, we created regression model of soil organic matter at the wavelength of 677, 1202, 2074 and 1509 nm and of soil moisture at the wavelength of 1432, 1524 and 1746 nm, employing multiple statistics. In addition, we used the regression model of soil organic matter to develop the map of soil organic matter concentration in the Yixing experimental area, employing imaging spectroscopy. Finally, we discussed the problems in the application of regression model.