Spatial Variability Of Fractal Dimensions For Soil Particle Size Distributions And Physical Properties

Soil particle size distribution influences many important soil physical properties largelyand integral soil particle size distribution curves can be used to speculate some important soiproperties which are unable to measure derictly or are costly to measure. Researches on thspatial variability of soil properties is basis of precision agriculture, and researches on thspatial variability of soil fractal dimensions for soil particle size distributions and relatedphysics properties can describe spatial influences on soil and crops more directly. There wermany researches on fractal dimension for soil particle size distributions, though, a few othem related about the spatial variability of it and related physics properties. More researcheshould be done deeply.A total of 582 soil samples were collected from field in Manasi in Xinjiang province ontwo profiles at 5-cm and 10-cm intervals respectively in this thesis, and soil particle sizdistribution, electrical conductivity, moisture and pH had been measured. The volumetrifractal dimensions (DV) and quantitive fractal dimensions (DN) for soil particle sizdistributions were calculated, relationships between fractal dimensions and soil physicproperties were discussed, and the spatial variability of fractal dimensions and physicproperties were analysed. At the same time, multifractal analysis was attempted using severaloess soils samples collected from Shaanxi province. There main research conclusions are afollows:(1) Researches on relationship between fractal dimensions for soil particle sizdistributions and soil properties showed that relationship between DV and clay contents wadistinct positive correlate. While relationship between DN and sand content showed distincnegative correlation.There was no direct quantify relationship between DV and DN. If the claycontent is larger and soil texture is minuter, the DV will be larger and DN will be smaller, sosoil particle size distributions'fractal dimensions can be used to estimate soil texturdiversities. Correlation analysises showed soil particle size distributions'fractal dimensionand particle contents had larger correlation, and most soil properties had moderate correlationwith electrical conductivities and soil water contents, however, electrical conductivities wernot so correlative with soil water contents.(2) Correlation analysises between soil electrical conductivities and other soil propertiein different directions showed that correlations were more distinct at vertical direction than horizontal at two profiles. Correlation between soil water repellency and particle content showed largest at 0.40~1.39μm, 26.3~144μm and 426~580μm, this result corroborated that sand content is one important cause of soil water repellency, as texture of sand content is rough.(3) Quantitive particle distribution characters and applicable ranges of four different kinds of loesses were analysed based on fractal theories. The results showed that fractal dimensions for quantitive particle size distributions obey to power law could only be calculated from limited granularity ranges. The lower limites of four kind soils were all above 0.5μm while upper limites were different distinctly.(4) Results of multifractal analysises of four kinds of soils showed D(q)'s variability of red–puddle soil were most obvious on whole q interval, this means stronger structure heterogeneity than others. The D(q)'s variability of sand-loess soil was the least. The curves ofτ( q)~q all showed concave and obviously different from lines of four kinds of soils, this means they have multifraction distribution. The curves of f (α)~αshowed protuberant functions, and the multifraction structure and dissymmetry degree of red-puddle soil were obvious than other three soils.(5). Soil properties'statistic value, directional semi-variogram functions, fractal dimensions based on semi-variogram functions and contour maps were obtained from researches on soil properties'spatial variabilities at two profiles. The statistic analysis results showed that range of DV was between 2 and 3, however, some of DN was large than 3. The variety ranges of soil properties at 5-cm interval profile were large than that at 10-cm interval. The means of soil properties at two profiles were similar. Semi-variogram functions could be modified by spherical models, exponential models, gauss models and linear models. Soil properties showed less directional features at 10-cm interval than those at 5-cm interval. The ranges of semi-variogram functions were smaller at 5-cm interval than those at 10-cm interval. The self-similar ranges of fractal dimensions based on semi-variogram functions were larger at 10-cm interval profile. There were no direct quantitive relation between semi-variogram function models'range and self-similar range, nor between fractal dimension and the parameter C0/(C0+C). Contour maps of soil particle size distributions'fractal dimension and particle content were silimar in trends, contour maps of properties related with soil texture showed three layers at 5-cm interval profile and two layers at 10-cm interval profile.