LANDSAT MSS data contains spectral reflectance information that is mainly about soil and vegetation except of course data from water covered areas. We want to have soil color information in order to recognise soil conditions and for soil mapping. Therefore, we have developed a method for extracting soil spectral reflectance information from LANDSAT MSS data, as follows. 1) We use a two dimensional graph of log IR and R for the determination of constants for use in the calculation of vegetation covering ratios. The G and R reflectances of plants are determined by IR: G and IR: R graphs. 2) When we need to skip some of the MSS data (compression data) so we pick up the most useful pixels from the data. Therefore, we select the pixels that are the lowest values in the IR band omitting those from water covered areas, and file the pixels of the R and G bands at the same location as the IR band. 3) We calculate vegetation covering ratios with the constants determined by method 1) and new filed data in method 2) . Then, we calculate the soil reflectance of the R and G bands using vegetation covering ratios, new filed data and R and G reflectances of plants. 4) B band reflectance is not discerned by LANDSAT MSS, so we interpolate it from R and G band data using a soil spectral model. 5) We make soil color images using R, G and B band reflectances of the soil. As to the application of this soil color image, we are able to extract the soil moisture changes between two LANDSAT scans. The method for accomplishing this is as follow. 1) Reconciling of data from two scans of the same location at different times. 2) Extraction of soil spectral reflectance. 3) Calculation of the changing value of soil reflectances between the two times. We can get useful information about soil distribution and moisture from this color image. We can understand the systems of using water on paddy fields with the image of soil moisture changes.