Study On Sampling Density In Soil Nutrient Spatial Variability Under The County Scale

In the study of spatial variability of soil nutrients, it was importance that the rationality and effective control of sampling density could meet the accuracy and could save resources. Taking Hengdong county Hunan province as example, using sampling and indoor analysis methods, according to the principle of traditional statistic and geostatistics, using variant function and spatial interpolation, based original data,10groups of different sampling density was selected, in order to identify the sampling density in county scale. And a reasonable sampling density was acquired through spatial interpolation of pH. Then all the points of soil nutrients were interpolated to original and reasonable sampling density. The results showed as follows:1.10groups of original data were selected and interpolated to pH. Correlation analyzing was performed between measured value and predicted value. And it shows that51point per1000hm2is the reasonable sampling density.2. Compared the interpolation of reasonable points to that of the original points, it showed that the mean value of pH of original points was5.31and that was5.33of reasonable points. Absolute error was0.02mg/kg while relative error was0.37%, which was in a reasonable range.3. The reasonable point density also could be used the interpolation of organic matter and available N. Compared to the interpolation of original points, the mean value of organic matter was31.9g/kg and32.3g/kg, respectivly. The mean value of available N was separate216mg/kg and213mg/kg, respectivly. The absolute error was3mg/kg while relative error was1.46%.4. The error of interpolation of available P and available K was relatively high using the reasonable point density, compared to the interpolation of original points. The relative error was under10%. The mean value of available P was17.9mg/kg and19.4mg/kg, respectivly. And the absolute error was2.5mg/kg and the relative error was7.82%. The mean value of available K was82.8mg/kg and91.6mg/kg, respectivly. The absolute error was8.8mg/kg and relative error was9.53%.