Effects Of Water Quality On Soil Water Characteristic Curve And Model Simulation In The Taklimakan Desert Highway Shelterbelt

Soil moisture is the main limiting factor for plant growth in arid regions.The Taklimakan Desert Highway Shelterbelt（TDHS）has been irrigated with 2-30 g/L saline water for a long time,which caused the salt accumulation on the soil surface.Thus,salt stress is another major problem faced by the shelterbelt.The shelterbelt construction significantly changed the soil physiochemical properties along the desert highway,and the research on soil water and salt dynamics will be helpful for sustainable management and utilization of the shelterbelt.The soil water characteristic curve（SWCC）is an essential hydrodynamic parameter in the study of soil moisture regulation and solute transport.In this paper,the soil of the TDHS was selected as the research object.Soil samples were taken from different soil layers（0–5 cm,5–10 cm,10–20 cm,and 20–30cm）and shelterbelt planted for different years（0a,16a,19a,and 25a）,bare land without vegetation cover among the shelterbelt（BL）were sampled.Shifting sandy land as control（CK）.All of the soil samples were treated with different salt types（Na2SO4,Ca Cl2,Na Cl）and solutions of different concentrations（0.5%,1%,2%,3%）.The effects of salt type and concentration and forest age as well as different soil layers on SWCC were systematically investigated,and the most suitable model was selected by fitting the commonly used models（van Genuchten model,Brooks-Corey model,and Gardner model）.The relationship between soil physicochemical properties and model parameters was further analyzed.On this basis,the transfer function of model parameters was established according to BP neural network.The main conclusions are as follows:（1）Soil physicochemical properties and SWCC w ERE significantly affected by the shelterbelt construction under saline irrigation,and such effect decreased with the increase of soil depth.Shelterbelt construction significantly reduced the bulk density of topsoil（0-5 cm）and increased soil EC and organic carbon content,and soil texture changed from sandy soil to loamy sand or sandy loam.Soil EC and organic carbon content of the CK increased with soil depth,while BL and US treatments showed an opposite trend.Compared with CK,the saturated water contentθ_s of BL and US increased by 4.42%and 12.67%,the field water capacityθ_fincreased by 68.9%and70.41%,and the available water contentθ_a increased by 87.84%and 70.97%,respectively.VG model is the most suitable for fitting SWCC of the TDHS soils.（2）Forest age,texture,salt type and solution concentration changed the SWCC trend and improved the soil water holding capacity.The soil water holding capacity increased with the increase of forest age,but the growth rate slowed down and ordered as 25a>19a>16a>0.The input of Na2SO4,Ca Cl2 and Na Cl improved the water-holding capacity of shifting sandy land and soils under the shelterbelt,and the promoting effect on the soil water holding capacity under the forest was Na2SO4≥Ca Cl2>Na Cl.However,the increase in water holding capacity of shifting sandy land caused by different salt types and solution concentrations was not significant.The effects of the three kinds of salts on the soil water holding capacity of under the shelterbelt were different,in which the greatest increase inθ_f andθ_r occurred for Na2SO4 and the greatest increase inθ_a was observed for Ca Cl2.Compared with the changes in salt and organic carbon,the response of soil water holding capacity to the texture was the most obvious.The change of soil water holding capacity in different soil particles ordered as FN（silt clay）>XS（fine sand）>ZS（medium sand）>CS（coarse sand）.When the soil particles changed from CS and ZS to XS and FN,the soil moisture constants increased significantly,and theθ_f,θ_a,andθ_ra increased by 6-9 times.（3）With the increase of EC and organic carbon,the fitting accuracy of VG model showed a decreasing trend,and the parameters a and n of VG model had a high correlation with soil p H,EC,organic matter,bulk density,sand content,silt content,and clay content.It was feasible to establish the soil transfer function of SWCC using soil bulk density,p H,EC,organic matter,sand content,and silt clay content as the input variables of BP neural network.In this study,the average relative errors of the predicted values of VG model parameters a and n by BP neural network were within 10%,thus,the established BP neural network prediction model was reliable and can realize the prediction of VG model parameters.In conclusion,the construction of shelterbelt with saline irrigation changed the physicochemical properties and water holding capacity of the shallow soils.The changes in salt,texture,and organic matter will altered the soil water holding capacity.The three kinds of salts showed a promoting effect on the soil water holding capacity under the shelterbelt.The VG model is the most suitable for fitting the SWCC of shelterbelt soils,and it is feasible to use BP neural network to correct its parameters.