Spatio-temporal Variability Of Soil Water Content On A Loessial Slope

Soil water is a main component of the terrestrial water resource. It participates in abroad variety of natural processes (hydrological, topographical, climatic, ecological)that act at different spatio-temporal scales. Change of soil water content is often thecenter of hydrological cycle and water balance. Sloping loessial land is widelydistributed on the Loess Plateau, it not only supports the rain-fed agriculture, but alsofunctions in places suffering soil erosion. Research on changes of soil water content atslope scale can help to understand the interrelations between soil water and vegetationand provide theoretical guides for soil and water conservation as well as vegetationrestoration. Also, such research would improve the effects assessment of land coverchange induced by Grain for Green Project on the Loess Plateau.This study was conducted on a loessial slope within the Liudaogou catchment ofShenmu county in Shaanxi Province, China. The study area is located in the transitionalbelt between the Loess Plateau and Mu Us desert. In2003, four adjacent experimentalplots (5m×61m) with different revegetation types (Korshinsk peashrub, i.e.Caragana korshinskii Kom., purple alfalfa, i.e. Medicago sativa, natural fallow and agrain crop of millet) were established on a uniform slope (12°-14°) with anorthwestern aspect. The four plots had relatively intact surfaces and were located awayfrom eroded gullies. To facilitate measurement of soil water content,11aluminum probeaccess tubes were installed at equal intervals of5m along the midline of each plot. InJuly from2010to2012,15measurement occasions of soil water content to a maximumdepth of580cm were taken. The data were collected to study the spatio-temporal variability of soil water content on a loessial slope. The main conclusions were showedas follow:1Soil moisture presented different vertical but similar horizontal trends in thefour vegetation type plots.For soils of the Loess Plateau, large changes in soil water content occur within thesoil depth of1m at monthly and seasonal scales. Also, a significant fraction of roots islocated in the first meter soil layer at all times. In chapter3, the distribution pattern ofsoil moisture at depths from10cm to100cm was investigated and the main findingswere:(1) Soil moisture in upper layers was higher than that in lower layers in both KOP(Korshinsk peashrub) and ALF (purple alfalfa) plots. In contrast, soil moisture trend inMIL (cropland with millet) was opposite and NAF (natural fallow) had no similar trendas the other three plots.(2) Soil was comparatively dry at deeper depth below the lowerparts of the KOP slope, and was moist at most depths in the counterpart of ALF plot.(3)Strong correlations in soil moisture for adjacent soil layers, while horizontal correlationwas widely observed among soil profiles on the slope.(4) The importance of factorsinfluencing soil moisture was ranked as: vegetation types, soil depth and slopepositions.2Temporal stability analysis showed that the most stable points were thosethat underestimated the mean SWS (soil water storage) of the plots and thetemporal stability was highest for ALF among all the four plots.Temporal stability of SWS within1m depth was analyzed (chapter4) using themethods proposed by Vachaud et al.(1985).(1) According to the frequency distributions,most measurement points did not maintain the same rank between the extreme soilwater conditions, especially for the points with probabilities of0.5.(2) Based on therelative difference analyses, the most stable points were those that underestimated themean SWS of the plots, and they were valuable for precisely estimating the mean SWSof the experimental plots, especially when corrected by an equation incorporating iandsi.(3) the plot-average points derived from different ways tended to be on themiddle or above middle slope in the plots, which provided a practical alternative torandom sampling intended to find the representative points of the areal mean soil watercondition.(4) Results of Spearman rank correlation coefficients indicated that the temporal stability was generally high for ALF, low for NAF, and lowest for MIL plotswhile for KOP plots it was intermediate. Revegetation types thus had significant effectson the temporal stability of SWS within the first meter of the soil profile. Revegetationtype in terms of its vegetation cover and aboveground biomass were the main factorsaffecting the temporal stability of soil water.3Allowing for soil moisture temporal stability in profiles, vegetation type, soildepth and the wetness index, in order of importance, had a significant effect on thetemporal stability of soil moisture.Further research on temporal stability varying with soil depth (chapter5) indicatedthat:(1) temporal stability of the soil moisture profile expressed by Spearman rankcorrelation coefficients generally increased with increasing depth in the KOP and MILplots, first increased and then decreased in the ALF plot and increased, but weresomewhat unstable on the first three measurement dates in the NAF plot.(2) Four typesof representative points, the driest, wettest, average moisture and most time-stable, forvarious soil depths in the four plots were identified by a relative-difference technique.Points with extreme moisture tended to remain representative at more depths than didpoints with average moisture and increased in temporal stability with increasing soildepth.(3) The correlation between MRDs and wetness index weakened with soil depth.In contrast, the relationship of SDRD to the wetness index varied nonlinearly with soildepth for all the plots. Vegetation type, soil depth and the wetness index, in order ofimportance, had a significant effect on the temporal stability of soil moisture. Amongselected soil properties, saturated hydraulic conductivity and bulk density significantlyaffected the MRD and SDRD. In addition, soil organic carbon had a significant effecton SDRD.(4) Temporal stability analysis enhanced the recognition of spatial pattern ofsoil moisture.4Merging the four plots as an integral slope, spatial variability of SWS wasmainly resulted from structural property, the spatial pattern of SWS in differentsoil layers were different to a large extent.Altogether44observation points in the four plots constitute a sampling grid of24m×60m. Using SWS data of three layers (0-100cm,100-200cm and200-300cm) inJuly of2010,2011and2012and methods of classical statistics and geo-statistics, chapter6reported the spatial variability of SWS. Results showed that:(1) the SWS wassignificantly different among soil layers and observation times, the SWS had moderatevariability, and (2) Gaussian models was mainly fitted for semi-variance of SWS. TheSWS had moderate spatial dependence; spatial variability of SWS was mainly resultedfrom structural property. The spatial autocorrelation distance decreased with decreasingsoil layers, ranging from14.7-18.4m.(4) Interpolation contour maps indicated SWSspatial pattern were very different among the investigated soil layers, which could beascribed to the different vegetation restoration measures on the small area. Comparisonamong different observation time indicated the dried areas expanded with time.5Soil moisture dynamics indicated a deteriorating water status underinfluences of vegetation restoration, time series of soil moisture at10cm depth washighly correlated to those in different depths for all the plots.Soil moisture data from10cm to340cm were employed to analyze its temporaldynamics (chapter7).(1) Soil moisture status in all the plots deteriorated with timeincreased, partly linked to the increase of potential evapotranspiration in thecorresponding periods for different year.(2) Time series of soil moisture at10cm depthwas highly correlated to those in different depths for all the plots. Soil moisture atsurface layer could be used to predict soil water content at other depths, i.e.20cm-70cm for KOP,20-90cm for ALF,20-50cm for NAF and20-70cm for MIL.(3) soilmoisture variability increased with increasing soil water content, however, which wasinterrupted by vegetation types, observational depths and time.6Soil drought aggravated in observational periods for soil moisture at amaximum depth of580cm. Furthermore, the thickness of developeddried-soil-layer (DSL) had been increased.The DSL is a special hydrological phenomenon on the Loess Plateau. Threeobservational points at a maximum depth of580cm in each plot were employed tostudy DSL formation characteristics and its development dynamics in chapter8.(1)According to the maximum depth of infiltration and water consumption, functionallayers of DSL in soil profile were classified into precipitation recharge layer, DSL andpotential DSL. The DSL in KOP and NAF plot ranged from100cm to580cm, from160cm to440cm, respectively. The upper boundary of DSL in ALF plot was120cm, its lower boundary was beyond580cm. The MIL plot had not developed a DSL due towater consumption was shallower than infiltration depth.(2) Field capacity in form ofvolumetric soil water content was used to determine the threshold value of DSL, and12%was recognized as the upper limit of DSL. Accordingly, the severity degree of DSLwas quantitively ranked.(3) Compared with the corresponding layers in2004, mean soilwater content for DSL in2010-2012decreased by44.5%,47.7%and17.5%for KOP,ALF and NAF, respectively. Soil moisture drop in MIL plot was far below the level inKOP and ALF, such result highlighted the dominant effects of vegetation in DSLformation processes.(4) Regarding MIL as a control, the evapotranspiration could besubtracted from other three plots. Since the plot established, SWS at every20cminterval in present DSL consumed by vegetation were8.82mm,9.68mm and2.58mmfor KOP, ALF and MIL, respectively. The ability of vegetation types to form DSL wasstrong for ALF, average for KOP and weak for NAF.(5) The status of DSL in soilprofiles worsen in2010-2012, the thickness of developed dried-soil-layer (DSL) hadbeen increased. Specifically, extremely severe DSL was formed in KOP plot, thicknessof severe and mediate DSL increased and reached to depth beyond our experiment,meanwhile, mediate DSL formed in NAF and its light DSL was in process ofdevelopment.