Spatial Variability Of Soil Properties And Anti-erosion Ability In A Small Watershed Based On GIS

Loess hilly region is one of the most severe areas of soil and water loss in China. Spatial variability of soil properties and anti-erosion ability are important factors for soil and water loss research. Spatial variability of soil properties research is a basis of application for soil erosion prediction model. Point data on soil properties could be translated into area data by spatial variance character and precise prediction maping, and this can enhance the precision of the application of prediction models. On the other hand, spatial variability of soil properties research is a basis of rationally utilizing the farmland, improving the soil and comprehensive control of water and soil conservation in the Loess Plateau.In this paper, method of traditional statistics and geostatistics combined with GIS was applied to determine and quantify the spatial variability and ordinary Kriging of soil properties and anti-erosion ability in the typical catchment. Furthermore, variance of soil properties under different vegetation were analysed and empirical equation of soil anti-scourability and saturated hydraulic conductivity were accomplished. The research results showed that:(1) The range of CV% of all soil properties is 5.34~62.22%, thereinto, capillary prosity and slit are belong to weak variability, the CV% of them are 8.54% and 5.34%, respectively. Others properties belonged to moderate variability, the CV% of non-capillary prosity is most with 62.22%. According to Skewness and Kolmogorov-Smirnov, data of soil slit, sand and crodibility do not fit normal distribution, soil oganic matter fit logarithm normal distribution, the others soil properties fit normal distribution except upward variables.(2) Under the assumption of the isotropy, the range of soil properties correlation distance was 74~186m, the maximum is field capacity and the minimum is soil aggregates. The nugget/sill ratio showed that weak spatial dependence for soil aggregate more than 75% and strong spatial dependence for erodibility less than 25%, and others exhibited moderate spatial variability. For different lag distance, the nugget/sill ratio of most properties was for enlarging the trend following lag distance. But that of soil bulk density and erodibility was marked change due to the deeper influence of construction factor. The range of soil properties correlation distance under different lag was 57~210m, the minimum was soil aggregate with 57~82m and the second was mean weight diameter(MWD) with 110~114m, and others were the range of about 130~170m.Under the assumption of the anisotropy, soil non-capillary prosity, field capacity, saturatedhydraulic conductivity, clay, slit, sand, aggregates, soil organic matter, soil anti-scourability and erodibility were typical geometric anisotropy, and others were zonal anisotropy. The conformity of the nugget/sill ratio of soil properties at different lag direction wasn't manifest, and if the nugget/sill ratio<50%, randomicity features was master soil variability, and if the nugget/sill ratio<50%, structural features was importance soil variability. That of correlation distance of soil organic matter and erodibility under the anisotropy was smaller than ones under the isotropy showed that decomposed angular intensified soil spatial auto-correlation. That of correlation distance of soil anti-scourability under the anisotropy was bigger than one under the isotropy showed that decomposed angular weakened soil spatial auto-correlation. Others were bigger or smaller simultaneity comparison with the anisotropy.(3)The range of fractal dimension of soil properties with the isotropy was 1.8386~ 1.9889, and the range of fractal dimension of soil properties with the anisotropy was 1.6629~ 1.9872. The fractal dimension of soil MWD and anti-scourability increased when the lag distance was lengthening, and the fractal dimension of saturated hydraulic conductivity, clay, bulk density, sand with 5 lags were nearer each other. The fractal dimensions of soil aggregates and non-capillary prosity with 4 directions were on the high side, and the ones of soil erodibility and organic matter were on the low side relatively. The fractal dimensions of sand, organic matter, MWD, erodibility and bulk density with different direction were displayed biggish discrepancy, and the fractal dimensions of soil aggregates, clay, slit, saturation with different direction were displayed less discrepancy.(4)Ordinarily Kriging based on structural analysis was applied to maping soil properities. The result of analysis with type indicated that vegetation is the most factor of spatial variability of soil properties and anti-erosion ability in the catchment. The distribution map of Kriging stardard deviation make known that the closer to sampling spot, the smaller stardard deviation, and the bigger on the other way round. The assessment of cross-validation reflecting that the accuracy of certain spatial interpolation of soil bulk density, saturated hydraulic conductivity, MWD, soil organic matter and erodibility was high relatively, whenas the accuracy of certain spatial interpolation of soil aggregates, capillary prosity and non-capillary prosity was low.(5)The rational sampling number for soil properties was given under different believable level and exactitude through the classic statistics analysis and its increased depended on accretion of coefficient of variation. On the condition of the same believable level as exactitude, the sampling number for non-capillary prosity, anti-scourability and saturated hydraulic conductivity was excessive, and the sampling number for bulk density, capillary prosity, clay and sand was little relatively. When P=95%, k=50% or P=90%, k=5%, the sampling number for non-capillary prosity, saturated hydraulic conductivity and anti-scourability was more than experimental sampling number, and others decreased largely than experimental sampling number. Sampling design with independence may be confirmed if the rational sampling number would be incorperated in correlation distance.(6)Soil properties under different vegetation were contrasted. In the Prunusd avidiana, Populus, Pinus davidiana, soil bulk density was smaller, prosity was abundant, water keeping and penetrating ability was better;the slit content in Pinus davidiana was much, the sand in valleyed sod was much;the soil aggregate and organic matter in valleyed sod and Prunusd avidiana were excessive;Soil anti-scourability and erodibility under all kinds of vegetation had certain discrepancy,the anti-erosion ability was strongest in the Prunusd avidiana for maximum anti-scourability and minimum erodibility, the anti-erosion ability was too bad in the Pinus davidiana for minimum anti-scourability and maximum erodibility. Except for aggregate and MWD, coefficient of variation in valleyed sod was most. Under uniform vegetation, the coefficient of variation for non-capillary prosity.anti-scourability and saturated hydraulic conductivity was big, and the coefficient of variation for soil bulk density, capillary prosity, clay and slit, erodibility was on the small side.(7)The nonlinear progressive regression method was adopted to found empirical equation for soil anti-scourability and saturated hydraulic conductivity respectively. The coefficient of determination carried through F test indicated that the equation was significant relation(a=0.01), and this model is of better precision and practical meaning.On the platform of GIS, the spatial variability of soil properties and anti-erosion ability was studied through variogram and fractal dimension. All these results showed that soil properties were regionalized variables, That is to say, they had the stochastic and structural characteristic together. Ordinarily Kriging based on structural analysis was applied to maping soil properities and anti-erosion ability and the stardard deviation of Kriging value was presented, the interpolation assessment of cross-validation illustrated that maping was provided with veracity. The conclusion of research made certain that vegetation as structural value was main factor for spatial variability of soil properties. Soil properties under different vegetation were contrasted, soil in Prunusd avidiana and Populus had good structure, and water penetrating and retaining capacity and anti-erosion ability were preferable.