Xiufeng WANG and Takashi INOUE （Hokkaido University）
This study investigates estimations of soil moisture on bare soil fields using soil line derived from satellite data. It was performed in two steps: (1)confirmation of soil line characteristics using ground measured data; and (2)comparisons of estimated soil moisture using soil line positions(first principal components) and some normalized type indices derived from satellite data. The soil line was confirmed by ground measured reflected red and near-infrared radiances (red and near-infrared reflectance) from soil; the first principal component(PC1)calculated from red and near-infrared reflectance (position on soil line) was sufficient to distinguish between moist and normal (or dry) bare soil surface or sparse vegetation surface. Using satellite data, the correlation coefficient and RMSE between the simulated soil moisture and PC1(position on soil line) calculated for any two wavebands from satellite data were better than those between the simulated soil moisture and the normalized type index(normalized index). We were able to estimate soil moisture accurately for a bare soil field using PC1 (position on soil line). The PC1 that was calculated by a green waveband with a middle-infrared waveband offered the best correlation coefficient and RMSE for estimation of soil moisture among some PC1s.
bare soil, first principal component, normalized index, satellite data, soil moisture