Soil Hydrological Characteristics Under Different Land Covers In The Upper Stream Of The Heihe River Watershed

The arid region in Northwest China covers 24.5%of the national territory and is home to over 26 million people.Mountain areas are the“water towers”of the arid area because of their importance in providing water resources for the downstream areas.Thus,knowledge about the soil hydrological processes is essential to understand the mechanisms of rainfall-runoff processes,and to improve hydrological modelling and water resources management in arid and semi-arid watersheds.However,the spatial heterogeneity of soil hydraulic properties and the mechanism of soil hydrological processes are still poorly understood in high elevation,complex-terrain mountainous areas,which restricts the parameterization and the test of the underlying mechanism of the hydrological models,and in turn impeding the hydrological modelling and water resources management in the mountainous areas.Accordingly,it is urgently needed to understandthe spatial heterogeneity of soil hydraulic properties and the soil hydrological processes in the mountainous area.To this end,this study aims to understand the soil hydrological responses under different land covers based on a soil moisture monitoring network in the Upper Reach of the Heihei River Watershed,Northwest China.The soil moisture monitoring network consists of 32 observation stations,monitoring soil moisture at profile of 0-70 cm with five layers?0-10,10-20,20-30,30-50,50-70 cm?during the growing seasons of 2014-2016 at a 30 min interval.First,soil samples under different land cover types:forestland,meadow,high coverage grassland?HCG?,medium coverage grassland?MCG?and barren land in the study area are sampled to measure the saturated hydraulic conductivity?K_S?.Vertical variation of K_S,in combination with precipitation intensity are used to analyze the dominant stormflow pathways of the profile under different land covers.Second,eight soil moisture stations with the same soil textural class and slope but different land covers?forestland,meadow,high coverage grassland?HCG?,medium coverage grassland?MCG?and barren land?are used to investigate the effects of different land covers on soil moisture response.Several indices are evaluated to quantitatively describe soil moisture dynamics based on soil wetting events.In addition,HYDRUS-1D simulation is used to further analyze the effect of land cover on soil moisture dynamics.Third,Cross-correlation analysis is conducted to analyze the coupling strength between surface?0-10 cm?and subsurface?10-20,20-30,30-50,50-70 cm,and profile of 0-70 cm?soil moisture.Afterwards,three different methods are tested to estimate subsurface soil moisture from in-situ surface observations:the exponential filter?ExpF?method,artificial neural networks?ANN?and cumulative distribution function?CDF?matching methods.The major findings are described below:1.The influence of land cover on the vertical variation of soil hydraulic propertiesResults show that K_S value in layer 5 is significantly lower than the values of above4 layers.K_S decreases in the order of forestland,meadow,MCG,HCG,and barren land,corresponding to the degree of vegetation degradation.The K_S decreases with depth under forestland,HCG and barren land,but increases first and then decreases under meadow and MCG.The dominant stormflow paths?DSP?for different land covers are different:forestland is dominated by deep percolation?DP?,HCG is dominated by subsurface flow?SSF?,meadow is prevailed by Hortonian overland flow?HOF?and has no SSF,while MCG and barren land are also dominated by HOF,but still form SSF.This result provides important information for improving the accuracy of mountainous hydrological modeling,and in turn leading to sustainable management of water resources in the study watershed.2.The response pattern of profile soil moisture to rainfall under different land coversOur results show that soil moisture response amplitudes along profile are similar under MCG and barren land,but significantly different under scrubland,meadow and HCG.The rate of soil moisture increment decreases significantly with depth for all land covers,except for the HCG.The temporal pattern of soil moisture increase is highly variable along the soil profiles depending on land cover type.In particular,the difference of response time between the adjacent layers vary from negative values to280 hours with depth.Preferential flow occurres mostly in soils covered by scrubland.Water transferability is higher in deeply rooted soil.Furthermore,sensitivity analysis indicates that soil hydraulic properties are key factors in regulating profile soil wetting events.Our results show that the soil moisture response indices are useful to quantitatively characterize patterns in profile soil moisture dynamics and provide new insights into the soil moisture profile wetting process?e.g.occurrence of preferential flow etc.?,which helps for effective model parameterization and validation,in turn improving hydrological modelling in arid high-altitude mountainous areas.3.The estimation of subsurface soil moisture from surface soil moistureCross-correlation analysis results indicate a strong coupling between surface and subsurface soil moisture in the study area,and the coupling strength decreases with the incease of depth.Both ANN and ExpF methods are able to provide accurate estimates of subsurface soil moisture at 10-20 cm,20-30 cm,and for the profile of 0-70 cm using surface?0-10cm?soil moisture only.Specifically,the ANN method has the lowest estimation error?RSR?of 0.42,0.62 and 0.49 for depths of 15 and 25 cm and profile SM,respectively,while the ExpF method best captures the temporal variation of subsurface soil moisture.Thus,ExpF method is further evaluated and applied with the SMAP_L3 product.Furthermore,results show that the area-generalized T_opt?optimum T?of the ExpF method in the study area can be used to estimate the subsurface soil moisture without significantly reducing the performance,when comparing to the station-specific T_opt.In a final step,the ExpF method is applied with SMAP_L3?Soil Moisture Active Passive?surface soil moisture products to estimate profile soil moisture,and the resulting profile soil moisture is compared to in situ observations.The results show that the SMAP-based profile soil moisture has a median Pearson's R of 0.68.The accurate of estimated profile soil moisture from the combination of SMAP_L3 product and ExpF method is significantly higher than the profile soil moisture generated from SMAP_L4and GLDAS.The SMAP profile SWI shows the spatial pattern of high SWI at east and low SWI at west of the study area,while the profile SWI increases from May to September then decreases to October.Overall,it is concluded that the ExpF method is able to estimate profile soil moisture from SMAP surface products in the Qilian Mountains.The results are important to the water resources management in the arid and semi-arid watershed.