| Vegetation restoration practices have been implemented to control soil and water loss in the Loess Plateau,but extensively afforestation has also produced some negative effects,such as soil desiccation.Soil moisture is a critical variable controlling hydrological and biological processes and hence plays an important role on vegetation restoration and ecological construction.Understanding the distribution and transformation characteristics of soil water on a typical slope in the loess gully region can therefore reveal the driving mechanisms of the water hydrological cycle and thus provide scientific evidence for efficient management of soil water resource and sustainable vegetation restoration.The 70 locations along two long transects were selected on a hillslope of the Loess Plateau,China.Also,eight experimental runoff plots(Bothriochloa ischaemum L.(BOI),Sea-buckthorn(SEB),Chinese pine(CHP),Chinese arbor-vitae(CHA),SEB+CHP,SEB+ Black locust(LOC),CHP+ LOC,and CHA+ LOC)were established on a natural steep slope(35°).Based on classical statistics,geostatistics,state-space approach,temporal stability analysis,water balance,and Hydrus-1D model methods,the objective of this study was to investigate the spatial-temporal evolutional patterns and temporal stability of soil water storage on a typical slope in the loess gully region,estimate soil water storage using a state-space approach,explore the distribution of actual annual and diary evapotranspiration along the hillslope,and to further quantify the changes of each component of a water balance model over time under eight different vegetation types.The main results are congregated as follows:(1)The soil water content exhibited a moderate variation at the hillslope.A significant difference was found among 0-1,1-2,and 2-3 m soil layers.The spatial pattern of soil water content along the hillslope was similar for 0-1,1-2,and 2-3 m soil layers,while soil water content of the 0-1 and 1-2 m layers had a greater coefficient of variation than that of 2-3 m layers.The soil water content of the 0-1,1-2,and 2-3 m soil layers were best fitted by Gaussian,Exponential,and Exponential semivariogram models,respectively(R2>0.90).A semivariogram of soil water content for 0-1 m soil layer indicated a moderate spatial dependence,while a strong spatial dependence was observed for 1-2 m and 2-3 m soil layers.Soil water content was significantly influenced by elevation,clay and silt content,bulk density,and leaf area index,and no statistical significant relationships were found between soil water content and sand content and saturated hydraulic conductivity.(2)Soil water storage had a strong spatial autocorrelation structure,and thus the sampling distance used in the present study for soil water storage was sufficient to identify their spatial representativity.At the 95% confidence level,soil water storage was positively cross-correlated with elevation,clay content,and fractal dimension and negatively cross-correlated with silt content.The state-space approach using any combination of variables described the spatial pattern of soil water storage much better than equivalent linear regression equations.The best state-space approach explained 98.8% of the total variation in soil water storage,while the best classical linear regression equation only explained 64.2%.Elevation and clay content were identified as the most effective combination for soil water storage estimation in the state-space approach,and were used to effectively predict the soil water storage spatial pattern along the second transect.The state-space approach is thus a useful tool that is recommended for predicting soil water storage spatial patterns at the hillslope scale using topography and soil properties.(3)Temporal variation of spatial mean soil water storage was most evident in the shallow(0-1 m)soil layer,while greater spatial variability was observed in the deepest(2-3)soil layer sampled.The temporal variation of spatial mean soil water storage decreases and the spatial variation of soil water storage increases with increasing soil depth.A strong temporal persistence in all soil layers is indicated by high Spearman's rank correlation coefficients(P < 0.01),with the temporal stability in deep soil layers relatively more stable than in shallow layers.Among four methods(MRD,mean relative difference;SDRD,standard deviation of relative difference;ITS,index of temporal stability;and MABE,mean absolute bias error)used to identify the MTSLs for the estimation of the mean soil water storage,the ITS provided the best results.The MTSLs identified in diverse layers of the first transect satisfactorily estimated the mean soil water storage(R2 > 0.90)and were used to effectively predict the mean soil water storage along the second transect(R2 > 0.91).Elevation,soil saturated hydraulic conductivity and leaf area index were the major factors influencing the temporal stability of soil water storage.The results of this study are useful for estimating mean soil water storage and should improve soil water management and hydrological applications in sloping black locust forested areas of the Loess Plateau.(4)The mean actual evapotranspiration of 35 locations at the hillslope was 424.4 and 474.2 mm in 2014 and 2015,respectively.The actual evapotranspiration showed a weak variation(CV=7.5%)in 2014,while a moderate variation was found in 2015(CV=10.4%).The actual evapotranspiration were both best fitted by Spherical semivariogram models(R2>0.95).Furthermore,a semivariogram of actual evapotranspiration indicated a moderate spatial dependence for both 2014 and 2015.The soil water content can be well predicted by Hydrus-1D model,and thus for daily actual evapotranspiration.The daily actual evapotranspiration in June-August was greater than that in May or September-October.The CVs of daily actual evapotranspiration for different locations were ranging from 47.79% to 56.28%.Actual evapotranspiration was negatively related to elevation,while it was positively related to sand content and leaf area index.No significance,however,was found between actual evapotranspiration and clay,silt content,bulk density,and saturated hydraulic conductivity.(5)Vegetation type substantially affected soil water storage and that the greatest soil water storage in both the shallow(0-2 m)and the deep soil layers(2-5 m)occurred under Bothriochloa ischaemum L.(BOI).Vegetation type also affected surface runoff and soil losses.The most effective vegetation types for reducing soil erosion were BOI and Sea-buckthorn(Hippophae rhamnoides L.),while Chinese arbor-vitae(Platycladus orientalis L.)and Chinese arbor-vitae(Platycladus orientalis L.)+ Black locust(Robinia pseudoacacia L.)were the most ineffective types.Soil water dynamics and evapotranspiration varied considerably among the different vegetation types.A soil water surplus was only found under BOI,while insufficient water replenishment existed under the other seven vegetation types.The higher water consumption rates of the seven vegetation types could result in soil desiccation,which could lead to severe water stresses that would adversely affect plant growth.This study suggested that both vegetation type and its effect on controlling soil erosion should be considered when implementing vegetation restoration and that BOI should be highly recommended for vegetation restoration on the steep slopes of the Loess Plateau.A similar approach to the one used in this study could be applied to other regions of the world confronted by the same problems of water scarcity along with the need for vegetation restoration. |
PDF Full Text Request