Scale Effect And Spatial Interpolation Models On Detecting Spatial Variability Of Soil Organic Matter In Different Geomorphology Types

The topsoil’s organic matter is not only indispensable nutrient element of plant growth, but also the necessary factor of agricultural land classification and farmland fertility assessment. Spatial variability of soil organic matter does exist objectively, In order to select a optimal sampling grid densities and spatial interpolation model, the paper study the spatial variation of soil organic matter in different scales and geomorphology types.Main contents of research:In this study, the hilly regions and coastal plain were choose as the research objective in geomorphology types and Zhangzhou as Prefecture-level city in typical areas and Longhai as County-level city in typical areas and Haicheng Town and Longjiao Town as Township-level in typical areas.According to different administrative regions, the grid densities of sample point were departed into four kinds. Grid densities of Prefecture-level city:2km×2km、4km×4km、6km×6km、8km×8km; grid densities of County-level city:0.5km×0.5km、1km×1km、2km×2km、4km ×4km;grid densities of Township-level:0.2km×0.2km、0.4km×0.4km、0.6km ×0.6km、0.8km×0.8km. Moreover, designing four different spatial interpolation models,these include KSOM、KSOI、KLUT and KLUS.There are 96 groups of sample data in different grid densities and spatial interpolation models. It then uses Kriging of Geostatistics to interpolate, after getting residual error, calculating the predictive value of the verification points. Taking RMSE that calcutated by formulae and Person correlation coefficient to gather statistics by SPSS as the standard of precision, finally we can draw the required sampling densities and the optimized spatial interpolation mode of characterize soil organic matter spatial variation in different scales and different geomorphology types.Results are as follows:1. Descriptive statistics in different geomorphology types:the coefficients of variation in hilly regions and coastal plain belong to the middle degree in different scales,and the coefficients of variation of Prefecture-level city and County-level city will gradually increase with the decrease of grid density and sampling points. In hilly regions, the coefficients of variation of Township-level tend to have no regularity. In coastal plain, the coefficients of variation fluctuate above and below 21% in Township-level. It is indicated that, when the research area is lesser in coastal plain,It has a smaller effect on sampling densities and the optimized spatial interpolation model of characterize soil organic matter spatial variation.In hilly regions, through the research of the spatial variation of all kinds of spatial interpolation method, we can draw the different required sampling densities and the different optimized spatial interpolation model, but the precondition is to ensure the soil sampling points combine soil types scientifically.(1)In Prefecture-level city, the required sampling densities and the optimized spatial interpolation model is KSOI combining soil types scientifically, grid density with 8km×8km.(2)In County-level city, the required sampling densities and the optimized spatial interpolation model is KLUS, grid density with 2km× 2km.(3)In Township-level, the required sampling densities and the optimized spatial interpolation model is KLUS, grid density with 2km× 2km, using KLUT, grid density with 0.2km×0.2km. but for grid density with 0.6km×0.6km, the amount of the sampling points is down by 40.30%,however, its precision just decrease by 0.42%. so, the KLUS is the best model and the most efficient grid density is 0.6km×0.6km.In coastal plain, there are differences between grid density and spatial interpolation model.(1) In Prefecture-level city, the required sampling densities and the optimized spatial interpolation model is KLUS, grid density with 8km× 8km.(2) In County-level city, The highest precision grid density is 0.5km ×0.5km, KSOI. If the accuracy demand is lower, using KLUS, grid density with2km× 2km. Compared it to KSOI, grid density with 0.5km×0.5km, the amount of the sampling points is down by 82.69%,but its precision just decrease by 14.02%.(3)In Township-level, the required sampling densities and the optimized spatial interpolation model is KSOI, grid density with 0.2km ×0.2km.