Quantitative Retrieval And Monitoring Of Water And Salt Using Remote Sensing In Arid Inland Zones

Water and salt are the important contents of arid inland research.Arid areas are often accompanied by drought.The most direct manifestation is the lack of soil moisture.Groundwater is the main source of water resources in arid regions.Due to unreasonable exploitation of groundwater and utilization of groundwater,the groundwater level changes drastically.Second,due to irrational irrigation and reclamation of arable land,soil salinization and secondary salinization have been caused.Therefore,it is very important to develop an objective,dynamic,and real-timely method for monitoring the soil moisture,groundwater depth and soil salinization.The purpose of this study was to develop a synergistic method for applying multi-sources remote sensing to retrieve the soil moisture,groundwater depth,and soil salinization in arid inland area(Yanqi Basin,Xinjiang,China)based on the Landsat-8 image,Sentinel-1A SAR data,MOD16 A2 data,and field measurements.The major accomplishments of this dissertation are summarized as following:(1)In order to explore the collaborative retrieval of soil moisture by optical and microwave remote sensing.The Water Cloud Model(WCM)was used to eliminate the effect of vegetation and calculate the microwave backscattering coefficient(?soil 0)based on the Sentinel-1A SAR data.The Modified Temperature Vegetation Drought Index(MTVDI)was calculated using the optical remote sensing data(Landsat-8)combined with the drought index model.Then,?soil 0 and MTVDI were applied to Support Vector Machine(SVM)regression algorithm in different combinations,and the adaptabilities of different combinations in soil moisture retrieval were discussed.The best accurate SVM model was achieved when considering MTVDI combined with ?soil 0 as the SVM model variable inputs.The modeling set decision coefficient(R2)and the validation set R2 of the model reached 0.81 and 0.89,respectively;the Root Mean Square Error(RMSE)of modeling set and the RMSE of the validation set were 3.16 and 3.15,correspondingly.(2)The soil moisture retrieval model was established using MTVDI and filed observation soil moisture data from Landsat-8 remotely sensed data.On this basin,based on investigation of groundwater level,soil moisture and other associated in situ data,the model of groundwater level distribution using remote sensing(GLDRS)was established.The groundwater depth was retrieved using this model and calculate the groundwater depth distribution.The validation of groundwater level retrieval model R2 and RMSE were 0.81 and 1.01,respectively.This study provides reference for monitoring groundwater level using remote sensing technology in arid oasis.(3)In this study,Back Propagation(BP)neural network model was employed as effective methods to monitor soil salinization.Soil Moisture(SM),Groundwater depth(GD),Salinity Index(SI)and Surface Evapotranspiration(SET)model parameters were used.SM was obtained from Sentinel-1A SAR and Landsat-8 data;GD and SI were calculated from Landsat-8 imagery;and SET was obtained from MOD 16 A2 data.This study showed that,the RMSE and R2 of the estimation obtained using BP neural network model were about 1.31 and 0.79,respectively,for the training data set;the RMSE and R2 were 2.60 and 0.72,respectively,for the testing data set.The results of this study indicated that BP neural network model has a great potential for estimating soil salinization using multiple-source remote sensing data.