The Research Of Optimal Sampling Scale Of Soil Nutrition Based On Different Landforms

Farmland provides the basic material goods for human beings, it is also the foundation of people’s daily life and manufacturing activities."Too many people, not enough land" is one of the China’s basic national conditions. In order to improve and make full use of the arable land, it is necessary to know accurately about the soil nutrient levels of arable land based on local conditions. At present, with the rapidly promotion of Soil Testing and Fertilizer Recommendation technology in China gradually, we also need to know the spatial and time variability of soil nutrients comprehensively and systematic. The quantized study of the spatial variability of soil nutrients can strengthen the forecasts and simulation of soil nutrients at different spatial scales, determine a reasonable scale or density of sampling, and improve the accuracy of soil survey and mapping. Under the circumstance of meeting the requirements of precision, the study of minimum number and rational layout of sampling points is particularly important. The Exploration of the spatial variability of soil nutrients applied in a reasonable sampling scale at different landforms partitions not only can save manpower, reduce sampling funds, but also a huge benefit for dynamic monitoring of soil nutrients which providing convenient access to real-time information on soil nutrients.In this study, geostatistics theory methods had been used to analyse the soil nutrient spatial variability. At the same time, take Wenxi County as the study area which has a typical landform and complex and varied terrain, it coexists plain area, hills with mountains. With the help DEM data, referring to the local terrain and cultivation methods, we divided the study area into three landforms:plain area’ area, hilly area mountainous area. We obtained soil nutrients data from cultivated land fertility evaluation, removed the irregular figure, and got the available samples.The number is3276. By using create fishnet function in ArcGIS9.3, we formed different grids of study area. Grid scale were100m×100m,200m×200m,300m×300m,400m×400m,500m×500m,600m×600m,700m×700m,800m×800m,900m×900m,1000m×1000m,1100m×1100m,1200m×1200m,1500m×1500m and2000m×2000m, a total of14scales. After meshing process, we intersected the sample points with the grid to obtain information on all the sample points in the grid, then took the samples that were near the geometric center of the grid to substitute the other points in the grid. Besides, GS+9.0and ArcGIS9.3had been used to fit the soil nutrients into the best semivariogram model under different scales of landforms partitioning and to interpolate soil nutrients by Kriging interpolation method. By cross-validation, we obtained the deviation between the estimated value and calculated values and interpolation accuracy. When the tolerance deviation was10%, we analyzed the optimal sampling scale of different topography partition. The results shows that the optimum sampling scale are plain area’area of900m×900m, hilly area of500m×500m.Soil nutrients in mountain areas have significant relationships with topography, slope, aspect,elevation and other topographical factors exists, Using the traditional statistical methods for typical sample collection according to the mountain landforms.