Estimation of soil moisture using pattern decomposition method and vegetation survey in Mongolian grasslands

In this study, we conducted the measurements of soil spectral reflectance and soil moisture content, and vegetation surveys over ten plots within the prefectures of Tov, Dundgovi and Henti of Mongolian grasslands. We attempted to apply a method for estimating soil moisture content to the field measured data. We also collected the data of vegetation species by the quadrate method, and calculated the vegetation cover ratio. The soil spectral reflectance of land surface was measured using a spectroradiometer which had a sensor in visible and near infrared wavelength (Wavelength range: 350 - 1050 nm). For an application to satellite imagery of optical sensors, four spectral bands were selected to be analyzed. The estimation method of soil moisture content by applying pattern decomposition has already been reported. In this investigation, we found that the soil spectral reflectance curve was steep when the soil moisture content is high, and becomes flat when the soil moisture content is low. Thus, in the pattern decomposition method, the flat model pattern is defined as the spectral reflectance pattern that corresponds to completely dried soil. The pattern decomposition coefficient for the flat model pattern was calculated from the pattern decomposition method using flat model pattern. The result of application of the suggested method to soil reflectance pattern from field survey showed a negative correlation (r**2=0.72) between soil water content and the pattern decomposition coefficient for the flat model pattern. This high correlation was shown throughout all the soil types in this study. The result of vegetation survey showed that there was a difference of soil water content between the two survey plots where low vegetation cover ratios were recorded. The reason for this result was assumed to be the effect of soil moisture deficiency or overgrazing.