The Study Of The Soil Spectral Haracteristic

Soil is the base of vegetation, from the remote sensing image interpretation point of view, soil is the background for vegetation interpretation. So the study on the character and variation of the soil spectrum is not only very important on the investigation of soil resource by means of remote sensing data, but also for the investigation of vegetation by way of remote sensing data, manpower visual interpretation, identification of automatic classification of computers. This thesis provided spectral properties of soil measured by ASD Field Spec Pro Dual VNIR Portable Field Spectrometer, and soil properties were retrieved from the soil reflectance.The thesis focused on retrieving soil properties from the hyper-spectral data. The hyper-spectral reflectance data were transformed to several mathematics mathematic transformations, such as reciprocal, logarithm, differential and so on. Using single-correlation analysis, the quantifying regression models between soil reflectance and soil particle diameter, and light height angle, and soil moisture were established.The main results are as follows, in the range of wavelength between 350nm～1050nm,1. Soil spectral characteristics in different particle sizes:The relationship of different layers of Eum-orthic anthrosols soil was that illuvial horizon > clay layer > overburden layer. Soil spectral reflectance had a negative correlation with soil particle diameter, and the correlation was strongest near 350nm. The differential (first order differential and second differential) of different transformations had no effects on detection of sensitive spectral index of soil particle fraction. Through testing, the exponentiation form of the linear regression model was the best in the overburden layer, while the fitting degree was not high in the most of wave band.2. Effect of the light elevation angle on soil reflectance: With the increase of the light height angle, the soil reflectance decreased first, and reached the minimum about 45°, and then increased gradually. Soil spectral reflectance had a weak linear correlation with the light height angle, and the strongest correlation coefficient was only 0.4. But the correlation coefficient increased when the reflectance and its reciprocal formation were transformed into the first order differential, and the highest correlation coefficient reached 0.7. There was a very good fitness of quadratic regression model of soil spectral reflectance and light height angle in each wave band, of which the correlation of determination was over 0.9. The reason why the soil reflectance was different in different light height angle could be resulted in the difference of shadow and roughness among soil particle.3. Effect of the soil moisture on soil spectral reflectance: The soil spectral reflectance had a negative correlation with soil water content. With the changes of water content, the soil spectral reflectance fluctuated in a small range when the soil water content was higher than 18%. There was a strong correlation between soil spectral reflectance and the soil water content, of which the correlation coefficient exceeded 0.8, and the longer the wavelength, the stronger correlation was. In the infrared part (760～1050nm), the fitness was very good of logarithm quantifying model of soil water content and soil spectral reflectance, of which the correlation of determination was over 0.9. So, we could adapt logarithm quantifying model to calculate the condition of the soil water content.