Responses Of Soil Moisture Estimated From CRNP To Vegetation Types In A Small Watershed On The Loess Plateau

Soil water is a principal component of Earth's water,playing an important role in the processes of energy exchange between the surface and the atmosphere and a variety of scale of water sports,and also is the main limiting factor for plant growth and development on the Loess Plateau.The spatiotemporal changes of soil water have important effects on soil erosion,coupled transport of water-heat-solutes,and material migration in the soil-vegetation-atmosphere continuum.The distribution and consumption intensity and depth of soil water are different due to different soil texture,vegetation types and allocation modes,which leads to the significant spatial differentiation of soil water on the Loess Plateau.In turn,the spatial and temporal distribution of soil water on the Loess Plateau determines the spatial distribution of vegetation types and their growth conditions.Therefore,the spatial and temporal distribution of soil water must be followed in order to ensure the sustainability of vegetation restoration.This study explores the changing process and distribution of soil water content?SWC?in 3 plots with different soil texture and different vegetation types on the Loess Plateau,so as to provide a scientific basis for the sustainable utilization of soil water in the Loess Plateau.The SWC and its distribution were measured by Cosmic-ray neutron probe?CRNP?and TDR sensors.The biomass and vegetation water content were measured by Unmanned Aerial Vehicle?UAV?.The changing process of biomass was modeled by SPACSYS model.The spatial and temporal distribution of SWC and its interaction with vegetation were analyzed and the main conclusions are as follows:?1?In 2016 and 2017,changing processes of SWC measured by the CRNP and TDR sensors in loamy-sand hill plot?LSH?,silty-loam dam plot?SLD?and sandy-loam slope plot?SLS?were consistent and the coefficient of the linear relationships between CRNP-SWC and TDR-SWC is the order:SLS>LSH>SLD.And the accuracy of the CRNP was verified by the oven-dried method(RMSE=0.068 cm³ cm^-3).CRNP-SWC shows a similar trend over time and is very sensitive to changes of SWC.The effects of soil texture on CRNP can be ignored.?2?The average SWC at a depth of 0?30 cm of three plots during the growing season was moderate variability and the order of varisions of SWC in 3 plots is SLS>LSH>SLD.The Spearman's rank correlation coefficient was low in the early period of the measurement and was high after July,indicating a stronger temporal stability of spatial patterns for 3 plots in the late period.And the correlation coefficient of each measurement date in the SLD sample plot is basically the highest value.The numbers of dry?MRD<0?and wet?MRD>0?spots in the three plots are approximately the same,but LSH and SLS deviate to a greater extent from the MRD=0 line,especially the wet points,indicating that wet spots in the two plots have a higher degree of deviation.The order of temporal stability of SWC in three plots is SLD>LSH>SLS.?3?Vegetation increases surface hydrogen content,which affects the CRNP measurement and the N₀ calibration method,so the N₀ can reduced RMSE of CRNP when the biomass at sampling date represents the average condition during the monitoring period.In this study,the most representative N₀ reduced RMSE from 0.121 cm³ cm^-3 to 0.068 cm³cm^-3.The representativeness of CRNP to regional average SWC decreases as the spatial variability of SWC increases.During rainfall,there are significant effects of vegetation on changes of CRNP-SWC.In the case of bare soil without rainfall,CRNP-SWC is very similar to TDR-SWC.?4?Three vegetation correction methods were used to calibrate CRNP-SWC and compared of three plots.The RMSEs between CRNP-SWC corrected by three vegetation correction methods and Kriging averaged SWC were reduced.The RMSE between CRNP-SWC corrected by Veg-N₀ in 3 plots are in the order SLS<LSH<SLD,and RMSEs of other methods in 3 plots,no matter with or without vegetation spatial distribution correction are order is LSH<SLS<SLD.Regardless of whether the vegetation spatial distribution is calibrated,the RMSE order is LSH<SLS<SLD.After calibration of the biomass spatial distribution,the RMSE of the three vegetation correction methods decreased,and the Veg-N₀ correction method had the lowest RMSE,0.031 cm³ cm^-3.Combining the CRNP with the UAV images can obtain a more accurate regional average vegetation distribution,which is a potential calibration improvement.Improving calibration methods that take into account SWC and vegetation distribution in the CRNP footprint should be useful for developing strategies for sustainable vegetation management in the Loess Plateau and other dry regions of the world.?5?The SPACSYS model simulated the water cycle,distribution and vegetation growth of Liudaogou watershed on the Loess Plateau well.The model was used to solve the shortcomings of obtaining biomass changes in long term by quadrat method and UAV method,such as vegetation destruction,poor representation,large workload and high cost.In this study,the model was used to simulate the SWC and vegetation growth in SLD and SLS plots.The RMSEs between the simulated and measured SWC in different depths were lower than 0.1 cm³ cm^-3,but the RMSEs between the simulated SWC and CRNP-SWC was higher than with TDR-SWC.The RMSEs between the simulated and measured SWC in deep layers?200?400 cm?were below 0.05 cm³ cm^-3.The simulated biomass was combined with the CRNP vegetation correction methods,and the RMSE between CRNP-SWC corrected by Veg-N₀ and Kriging averaged SWC was 0.071 cm³ cm^-3.Although its RMSE was higher than the results of UAV biomass,considering the advantages of long-term daily changes in biomass,low cost,and low workload,model simulation can be used to obtain and predict biomass changes on the Loess Plateau.?6?The changes of soil moisture of 0?400 cm under several typical vegetation covers were analyzed,it was found that SWC changes sharply in the shallow layer?0?50 cm?but basically showed as grassland>farmland>shrub>tree,while SWC in the deep layer?50?400 cm?was deficient except for farmland.The differences of soil texture between SLD and LSH resulted in the difference in SWC of caragana at different depths.Points measuring methods,such as TDR,cannot represent the regional average SWC,indicating the necessity of the application of CRNP.But the high biomass and the difference of spatial distribution of vegetation lead to the error of SWC measured by CRNP,which reduce the representativeness of CRNP-SWC to the regional average SWC,though the representativeness of CRNP is still higher than TDR.It is recommended that the planting density in LSH and SLD should be reduced appropriately,or the amount of high water consumption vegetation should be reduced.The current vegetation and planting density of SLS are relatively reasonable.