| Understanding the spatial heterogeneity and distribution patterns of soil physicalproperties are fundenmental for various corresponding researches. Under the natural andanthropogenic influences, landscape pattern is complex in the middle basin of the HeiheRiver. Soil properties differ significantly both vertically and horizontally. Considering thecomplex landscape pattern characterized by the coexistence and interaction of desert andoasis, this study focused on the oasis and oasis-desert ecotone. Soil water contents to adepth of3m in the120sampling points in the100km2study area were measuredregularly in2011and2012. Soil organic carbon concentration, hydraulic parameters andmechanical composition in the respective soil layers were measured. By using theclassical statistics and the one dimensional Markov chain theory, the transitioncharacteristics of different textural layers in the vertical and horizontal directions wereanalyzed, and the three-dimensional visualization of soil textural profiles was realized.Soil organic carbon stocks in the0-3m profiles in different landscapes were estimated,and the spatial variability and distribution were analyzed. The temporal stability of soilwater storage in the profile of different landscapes was analyzed and an a priori approachto identify the representative locations was proposed. The temporal variation ofcomponents of soil water balance was analyzed in the irrigated cropland during the thegrowing seasons of summer maize. Efforts were made to accurately estimate the spatialmean soil water storage and spatial mean deep percolation. The spatial variability ofhydraulic parameters was also analyzed. The main results obtained were listed as follows:（1）There were seven textural types in the study area, namely, sand, loamy sand,sandy loam, loam, clay loam, silty clay loam and silt clay, respectively. Compared with the non-occurrence of silty clay layers in the surface soil, another six typesof textural layers all occurred in the surface soil, while sand layers occurred witha much higher probability than others. The layer thickness of each textural typecould be characterized as a lognormal distribution, with relatively thicker sandand silty clay loam layers, and relatively thinner sandy loam and clay loam layers.For a certain soil type in profile, layers occurred beneath it were mainly the twowhose textural types were similar to it, especially the one consisting of more fineparticles. The formation of loamy sand layers was much strongly dependent onthe lower layers, whereas clay loam layers had a key effect on the formation ofthe upper layers. Loamy sand and loam layers had relatively high probability toown sandy loam layers as upper layers, while silty clay loam layers had relativelyhigh probability to occur as upper layers of both clay loam and silty clay layers.None of the seven textural types had the same combinations of upper and lowerlayers simultaneously. Markov characteristic and the stability of the verticalchange of adjoined textural layers were verified. One-dimensional embeddedMarkov chain model could accurately describe the vertical change of soil texturallayers. The main combinations of textural layers were sand-loam, loam-sand,loam-clay, and clay-loam.（2）The one-dimensional continuous Markov chain model constructed based on thetransition probability matrix in each of the principal x, y and z directions reflectedthe spatial variability of soil textural layers in each direction of the study area. Themean lens length, mean proportion and the entropy factors quantitively describedthe characteristics of the distribution of, and transition among different texturallayers. The distribution of soil textural layers were not completely random, buthad the juxtapositioning tendencies. The mean lens length ratio was used to obtainthe lens length in the horizontal direction according to the lens length in thevertical direction. Then the transition rate matrix was calculated and theone-dimensional continuous Markov chain model was constructed in thehorizontal direction. This procedure solved the scarcity of data in the horizontaldirection to some extent. The one-dimensional continuous Markov chain models in the the principal directions were interpolated into a three-dimensional Markovchain model in an arbitrary direction, which reflected the cross co-correlationsamong different textural layers, and reproduced the soil textural profiles in thestudy area. By integrating the geostatistical information in the simulation, theMarkov chain model reflected the continuity in space, the asymmetry and theanisotropy of the distribution of soil textural layers in the study area.（3）Soil organic carbon (SOC) density was low and remained homogeneous in theprofiles of desert soil. The vertical distributions of SOC density in cropland andwetland could be well described by logarithmic functions. Soil organic carbondensity presented moderate spatial variability and strong spatial dependenceacross all depths. Wetland and desert could be easily recognized by the highestand lowest SOC densities in the study area, respectively. Soil organic carbondensities in the3-m profiles were59.4,149.6and174.4Mg ha-1for desert,cropland and wetland, respectively, among which, about67.0,52.7and58.0%were stored in the1-3m layer, respectively. Clay plus silt (clay+silt) particleswere the major determinant of SOC in the study area. The variability in SOCdensity explained by clay+silt content, increased with depth in desert andcropland, but decreased with depth in wetland. The remaining SOC densityvariability could be attributed to factors not included in this study, such asgeography, vegetation and the degree of erosion. Errors in the measurement ofSOC concentration and the distribution of soil-particle size, however, mayintroduce uncertainty in the determination of soil bulk density and thus theestimation of SOC density. The concentration of SOC in the0-0.3m layerincreased by196.3%after the reclamation of native desert for less than40yearsand decreased by5.3%after the cultivation of wetland as cropland for less than30years. Short-term cultivation was insufficient to significantly alter SOCconcentration in the deeper layers of desert and wetland soils.（4）Spearman’s rank correlation coefficients and the relative difference analysis bothindicated the temporal stability of soil water storage (SWS) in profiles of desert,cropland and wetland, respectively. The temporal stability of SWS spatial patternincreased with depth for desert and wetland. The representative location was identified as the one with the smallest standard deviation of relative differences.No single location could represent the spatial mean SWS of the three layerssimultaneously for each landscape. Soil texture and soil organic carbon contentcould explain much of the temporal variability of SWS spatial pattern. At therepresentative locations in the three layers of desert, the cumulative probabilitiesfor clay, silt, sand and soil organic carbon contents were <0.25,<0.25,>0.75and between0.5and0.75, respectively, and the respective values in the croplandwere>0.75, between0.5and0.75, between0.25and0.5and between0.5and0.75. An a priori approach was then proposed to select the potential representativelocations more conveniently in larger areas of the desert and cropland, from whichactual representative locations can be identified after long period measurement.（5）Although the growing stages of summer maize, the amount and times of irrigationall differed in the northern and southern croplands in the study area, soil waterstorages in the northern and southern croplands were temporally stable. Thelocation with the lowest standard deviation of relative differences couldaccurately estimate the spatial mean SWS with high coefficient of determination(R2>0.91, P<0.001) and prediction accuracy (PE>0.76) and near-zero meanabsolute relative error (MARE=0). From May21to late September in the twoyears, about39%and22%of the irrigation and rainfall lose as deep percolation inthe northern and southern croplands, respectively. Deep percolation at therepresentative location of the0-1m spatial mean SWS could generally estimatethe spatial mean deep percolation in the2m soil profiles of the northern andsouthern croplands.（6）In the0-10,10-20and20-30cm soil layers, saturated water content exhibitdmoderately spatial variability, and hydraulic conductivity and bulk densityshowed strong variability over space. Wetland and desert had the highest and thelowest saturated water content, respectively. Hydraulic conductivity in desert wasthe largest. Saturated water content, bulk density and hydraulic conductivity allshowed moderatedly spatial dependence. Saturated water content and hydraulicconductivity had the largest and smallest ranges over space, respectively. The formula of soil water characteristic curve proposed by van Genuchten fitted themeasured curves with high coefficinets of determination. In desert, the volumetricsoil water content drastically declined when the soil water sunction was high, andkept nearly constant when the sunction was lower than a certain value. Soil waterretension curves in wetland were relatively gentle. The measured soil waterretension curves at various locations could be represented by a uniformed curvefor each landscape after scaling.Insight into the soil textural profiles and the spatial variability of hydraulicparameters can benefit studies on the processes of water movement and solute transfer instratified soils. Knowledge about soil organic carbon stock, its spatial variability anddistribution, and the influencing factors may help to make measures to restore thedesertified land. Basedon the accurate estimation of the carbon stock potentialities ofdifferent landscapes, land use and landscape layout can be planed and managedreasonably in the hope of playing the potential carbon sink role of arid region. Analyzingthe temporal variation of soil moisture in different landscapes and estimating thetemporal variations of soil water balance components in irrigated cropland during thegrowing season of summer maize are fundemental for further study on the soil waterprocesses in the SPAC system of the oasis ecosystem in the middle basin of the HeiheRiver, and favour for the improvement of irrigation water use efficiency and thesustainability of oasis ecosystem in inner arid region, northwestern China. |
PDF Full Text Request