Spatial Variability And Scale Effects Of Soil Water And Salt Of Upstream In Heihe Basin

In this paper,based on the farmland soil of Yaozhuang in Minle county，chestnut andsandy soil in different scales (the scale one is one kilometer spaced of mountain soil samples,the scale two is50meters spaced of farmland soil samples, the scale three is3meters spacedof farmland soil samples) in Sunan county of upstream in Heihe basin in Gansu province,traditional statistics, geostatistics, autocorrelation analysis, fractal dimension model,multifractal theory, wavelet analysis were applied to research the spatial variability of soilproperties such as moisture content (θ), conductivity (EC), organic matter (OM),clay (CL),silt (SI) and sand (SA)by combining with software of GS+, DPS, Surfer and Matlab, toprovide a scientific basis for the management of agricultural soil. The main conclusions wereas follows:(1) Traditional statistics analysis soil properties spatial variability of different scales,different directions and different types of mountain.The results show that: θ and EC of scaleone be in a strong degree of variation, OM, CL and SA be in a moderate level of variation, SIbe in a weaker degree of variation. SI of scale two and three also be in a weaker degree ofvariation, other properties be in a moderate level of variation.In addition to EC which has astrong degree of variation in the east-west direction, other farmland soil properties have amoderate level of variation.EC of chestnut soil, θ and EC of sandy soil have a strong degreeof variation. In farmland soil samples, θ always has a significant correlation with CL, SI andSA.θ for chestnut soil samples significantly correlate with EC, CL and SA, while θ onlycorrelates with OM and EC in sandy soils.(2) Geostatistics analysis θ, EC, OM, CL, SI and SA of farmland soil, the results showthat: EC for grid soil samples, EC and θ for transect soil samples have a strong degree ofspatial variation. This was mainly affected by random factors.CL of transect soil samples inthe northwest to southeast direction was in a weak degree of spatial variation which wasmainly attributed to structural factors. The other soil properties were in a moderate level ofspatial variation because of double effect by structural and random factors.(3) Spatial autocorrelation method analyzed autocorrelation of θ, EC, OM, CL, SI and SA of farmland soil with the change of distance. Results show that the maximum correlationlocation is inconsistent. Cluster or discrete distribution of soil properties is not obvious tooccur.(4) Fractal dimension model analyzed soil particle volume fraction of farmland soil,chestnut soil and sandy soil of upstream in Heihe basin. Results show that the higher thefractal dimension, the more heavy clay soil texture, conversely the more loose.(5) Multifractal theory analyzed spatial variability of farmland soil properties in differentdirections.The results indicated the following statements: curves of logμ(q,δ)-logδ, τ(q)andD(q) revealed θ, EC, OM, CL, SI and SA have multifractal structure. θ and EC are especiallystronger. Although multifractal spectrum of soil properties in different directions are all clockshaped, opening size and symmetry are different. The degrees of asymmetry and uneven of θand EC are relatively larger than other soil properties. The Δα of θ in the east-west direction is1.05which has the maximum degree of variation. The Δf in the following situations are lowerthan zero: θ in the east-west direction, SA in the northeast-southwest, EC in all four directionsand OM in all other directions except for the southeast-northwest direction. Moreover, thefractal spectrum curve trails to the left side, which means that the medium and high value ofsoil property play a leading role in spatial variability.(6) Wavelet method analyzed spatial variability of farmland soil properties in differentdirections.The results shows that: with the changing of distance, wavelet variance curves ofsoil properties in different directions present a fluctuated rule and have oscillator periods. Thelarger the amplitude is, the greater variation is. Different spatial scale corresponds to differentstructural changes in soil properties. Changes in the small-scale in nested larger scale show amore complex spatial variability structure.The amplitude of θ and EC is larger than other soilproperties in the east-west direction, which illustrated that there is greater degree of variation.Wavelet variance curves of EC have more oscillator periods and plurality of scale structures,which shows that the spatial variability is more random and independent.The spatialvariability of OM in the northwest-southeast and southwest-northeast direction is strongerthan that in the south-north and east-west direction.The spatial variability of CL, SI and SA inthe southwest-northeast directionis stronger than that in the northwest-southeast direction.(7)Geostatistics, multifractal theory and wavelet method all analyzed the spatialvariability of θ, EC, OM, CL, SI and SA of transect soil soil. Conclusions are substantiallyuniform. In certain scale, θ and EC have stronger spatial variability than OM, CL, SI and SAin different direction, especially in the east-west direction.