The Spatio And Temporal Variation Of Land Use/Cover And Soil Salinity In Changing Environment In Shenwu Irrigation District

Aiming at water saving project in Shenwu Irrigation area of Hetao Irrigation District in Inner Mongolia,the paper started with land use/cover type and soil salinity these two parts,applying remote sensing and geographic information system,and also using models like markov model and back propagation neural network model,combining with classical statistics knowledge,providing some basic data for water right conversion project as well as providing some scientific proofs for land use/cover and change of soil salinity conditions after water rights conversion project has been implemented.The main conclusion of this paper is as follows:1.Choosing remote sensing image in different time periods cooperating with different band combination can assist artificial perception to recognize different land-use types.Using NDVI,MNDWI,SI index as the characteristics of the decision tree algorithm.This can effectively enhance the classification accuracy,total classification accuracy and Kappa coefficient come to 91.25%and 0.89.2.The area of sand and saline in the 20 research area decreased by 43.78%and 62.50%respectively,Natural vegetation,vegetation,water and other areas increased by 135.16%,75.08%,42.71%and 29.41%respectively,The area of sand reduction is mainly due to the conversion of natural vegetation and vegetation,The reduction in the saline is mainly due to the conversion of vegetation.3.After verifying,using Markov model for predicting land use/cover type is viable,in the condition of immovable driving force,the sand area will decrease to 410.56 km2 till 2035;The area of natural vegetation and saline tends to be stable,keeping 400 km2 and 94 km2 respectively;The area of vegetation and water keeping increased to 941.16 km2 and 152.44 km2 respectively.4.Using the SI index,Adbole and band 7 of Landsat OLI as characteristic quantity,the decision tree classification of the decision tree threshold was determined by mathematical analysis to determine the extent of the different salinization fields,total classification accuracy is 79%and Kappa coefficient is 0.72.The research areas of nonsaline land,mild salinization of farmland,moderate salinization and heavy salinization of farmland constitute 33%,41%,19%and 7%of farmland respectively.5.Using the BP Ann model to simulate the relationship model between the monitoring point EC and the various land utilization types,when the input factor is Dchannel,Dsaline,Dsand,Dvegetation and Dother,nerve cells in hidden layer are six,correlation index R is 0.94,RMSE is 0.16,the average error validated is 21.66%.The simulation effect of model is better than multiple linear regression model and other back propagation model.6.With the implementation of the water-saving renovation project,the groundwater depth in the water-saving area in September 2016 increased significantly compared with that in the year of September 2015.The soil salinity of the 0-10cm soil water-saving area in the water-saving transformation area showed a significant decrease The correlation between groundwater depth and soil salinity was lower than that in the study area.