Applicability of Estimation Method for Soil Moisture in Mongolian Grasslands using a Pattern Decomposition Method

抄録

The objective of this research is to develop a method that will allow remotely sensed data to be used for estimating soil moisture and distinguishing vegetated and bare ground areas in Mongolia. The study was conducted in central Mongolia, where climate conditions and human impacts are resulting in degradation of grassland and damage to grazing stock. The method employed was a pattern decomposition method using flat pattern model for estimation soil moisture, which is developed by authors. A total of ten survey plots were established in grasslands, and a spectroradiometer was used to measure spectral characteristics of the soil in the visible and near-infrared wavelengths. Spectral reflectance of the soil was measured, and used to derive the pattern decomposition coefficient and estimate soil moisture. These results were then compared to actual field measurements of soil moisture to verify the accuracy of the method. In addition, photographic images were used to estimate the vegetation coverage for each survey plot, and the effects on vegetation on the accuracy of soil moisture were examined. The results showed a strong negative correlation between water content in soil and the pattern decomposition coefficient for flat model, indicating that this method is capable of accurately estimating soil moisture regardless of soil type. A strong correlation was also found between the vegetation coverage and the pattern decomposition coefficient for vegetation, suggesting that the method can be used to estimate vegetation coverage. A high level of accuracy, however, was achieved only for bare ground, indicating that the vegetation reflectance influences the spectral reflectance of the soil, and the current method is thus not applicable to vegetated areas.