Quatitative Estimation Of Shallow Groundwater Level Using SVM In Oasis-desert Ecotone Of Ugan-kuqa Region

Water is a necessary resource for living things.The issue of sustainab le development of water resources is a basic scientific issue facing the world.Groundwater has a wide range of distribution,and its sustainable use is strong.Therefore,it occupies an important role in water resources and is essential for the construction of ecological civilization and sustainable development.Reasonable shal ow groundwater levels can meet the normal growth needs of oasis vegetation in arid regions.This makes it particularly important to study the interaction between shal ow groundwater,vegetation and soil,and the distribution of vegetatio n communities and shal ow groundwater.For arid regions,precipitation is scarce and evaporation is strong.Groundwater is the main mode of water supply and the ecologica l environment is fragile.Especially,the role of the oasis ecological environment in arid regions is particularly important.The in-depth study of shal ow groundwater,vegetation,and soil interactions has important implications for the stable developme nt of the ecosystem,and also determines the process of oasis and desertification that are opposed to each other in arid regions.In this study,Xinjiang Ugan-Kuqu Oasis was selected as the research area.Quantitative inversion was performed using microwave and optical remote sensing data,soil moisture and groundwater depth data,combined with vegetation and soil conditio ns.By the survey of soil moisture and groundwater depth in the study area.Study draw several conclusion as follows.?1?Directly inversion of groundwater depth through Ts-VI feature space model.?2?Based on the support vector machine regression algorit hm in combination with microwave and optical remote sensing data,a soil moisture inversion model was established and groundwater depth data was fitted.?3?The results show that the groundwater depth information can be obtained from two differe nt inversion models,and the depth of groundwater indirectly inferred by SVM is better,and can better meet the actual situation in the study area.The main findings and results of this study are as follows:?1?This study directly inverts the data of groundwater depth through the TVDI model.Based on the soil samples collected in the field,we processed and analyzed the soil samples.We found that the correlation between the soil depth at different depths and the groundwater depth was different.After analysis,the correlation between 0-10cm soil water content and groundwater depth was the highest,followed by 40-Soil moisture content of 60cm soil layer,10-20cm and 20-40cm soil layer phase relations hip is relatively low.This conclusion provides data support for inversion of soil moisture in later period and inversion of groundwater depth.?2?Using the C topographic correction model to correct the surface temperature,and different The vegetation index and terrain corrected temperature data were used to construct four different T_S-VI feature spaces,and the correlation between T_S-VI feature space data and groundwater depth data was analyzed.Studies have shown that the Ts-MSAVI and Ts-EVI feature spaces can better reflect the depth of groundwater depth in the spring of the study area.Based on this,the groundwater depth is inverted to obta in the spatial distribution of groundwater level.?3?In this study,The TVDI model parameters and backscatter coefficie nt parameters of different vegetation indices were established,and the soil moisture was inversed to the SVM regression model based on the parameters that were used together.Through experiments,we know that TVDI_MSAVI achieves the best accuracy,followed by TVDI_EVI.When considering only the?⁰_soil parameters of the model,the modeling result R² is 0.64 and the verification result R² is 0.70.Using soil moisture?⁰_soil and TVDI as parameters of the SVM model for soil moisture inversion,?⁰_soiland TVDI_MSAVISAVI model the best accuracy.Compared with considering TVDI and?⁰_soiloil parameters alone,the accuracy of the model established based on the common parameters of TVDI and?⁰_soil has been significantly improved.Compared with TVDI only,the R² of the modeling set increased by 0.12,0.12,0.11,and 0.12,respective ly;The accuracy result is best obtained by inverting groundwater depth in soil moisture to obtain groundwater depth in the study area.?4?By comparing the two,it can be concluded that the accuracy of groundwater burial depth of the support vector machine regression algorithm Ts-MSAVI and microwave indirect inversion is relatively high.