Use of disjunctive cokriging to estimate soil organic matter from Landsat Thematic Mapper image
In Japan, it has recently been decided to grow crops other than paddy rice in the paddy fields, since the demand for rice has decreased. This study was carried out in order to obtain information about this decision. Since soil organic matter is important for soil fertilities, detailed descriptions of the spatial variability of soil organic matter, therefore are desirable in this decision. The spatial variability in the Aidu Basin, northern Japan, was predicted using disjunctive cokriging. The disjunctive cokriging estimator used soil sample data as the variate, and as covariate, estimates of soil organic matter obtained from Landsat Thematic Mapper images of submerged paddies. The disjunctive cokriging estimator provided a better estimate than a traditional regression analysis. However, we should pay attention in using the estimate. Since soil organic matter content might not be directly linked to crop production for some soils, we should make these decisions by comparing the estimate with conventional soil maps. This paper also investigated relations between the spatial structure of soil components, which have a major influence on soil colour, such as organic matter and that of each TM band data, using semivariograms. The nested structures of semivariograms for soil colour components could be found in semivariograms from TM Bands 1-3. The nugget variance and range in the band data were influenced by spatial patterns of soil colour components.