| Under the background of global change,the contradiction between population growth and resource shortage has become increasingly prominent,leading to an increase in the intensive use of land resources,and the spatial pattern of land use is in a dynamic evolution.At the same time,the problems of land degradation such as soil salinization pose new challenges to the sustainable use of land resources.Therefore,developing analysis methods of land use and soil salinity evolution and exploring the evolution characteristics under historical and future development scenarios,not only can elaborately understand the basic attributes and evolution characteristics of land use pattern and salinization,but also provide the basic support for comprehensive protection of land resources under changing environment.Taking Hetao Irrigation District(HID)in Inner Mongolia as an example,the spatiotemporal evolution of land use from 1986 to 2016 was analyzed using the maximum likelihood supervised classification method.On this basis,the surface soil salinity inversion models for farmland were established using the regression analysis and spectral feature space theory,respectively.The best model was selected through accuracy assessment and applied to the research on spatiotemporal evolution of farmland soil salinization.Based on the 30-year historical evolution analysis of both the land use and soil salinization,the CLUE-S land optimization configuration model was utilized to predict the soil salinization patterns of the next 30 years.In the CLUE-S model,each degree of farmland salinity was regarded as an independent land type.The soil salinization pattern predictions were also under several scenarios,such as planting structure adjustments and enhancing water-saving intensity.Then,a comprehensive index was established to evaluate the degree of soil salinization in the study area.Accordingly,the proper layout of crops planting structure matching the spatial pattern of soil salinization in HID was recommended.Conclusions were as follows:(1)Based on the spectral feature space theory,the accuracy of the farmland salinity inversion model combining a vegetation index with a soil salinity index was significantly improved compared with the regression models.The model that combined Modified Soil Adjusted Vegetation Index(MSAVI)with Soil Index(SI3)showed the highest accuracy for salinity inversion on farmland surface soil(R2=0.76,RMSE=0.30dS/m).Then,the model was applied in estimating the farmland salinity in 2016 and obtained a satisfied result.The results proved the rationality of this method in the study of soil salinization in arid and semiarid irrigation areas.(2)The spatiotemporal evolution analysis from 1986 to 2016 showed that various types of land use and salinization show a cross-distribution pattern in HID,and farmland occupied most of the study area.The degrees of farmland salinization in HID were mainly non-saline and slightly saline,and the eastern and the northern part were more serious than the western and the southern part of HID.Salty barren land mainly located on the edge of Wulan Buh Desert,the central and the northeastern part of HID.The largest expansion of land use type was non-saline farmland,followed by slightly saline farmland,grassland,and construction land.Extremely saline farmland and woodland had smaller expansion.The largest decrease in area was unused land,followed by salty barren land and water.The areas of moderately and severely saline farmland decreased a little.Overall,the soil salinization in HID showed a trend of decrease.The average surface soil salinity decreased by 22.45%from 1986 to 2016.Especially since the implementation of water-saving irrigation,the surface soil salinity remained relatively stable and low.(3)The CLUE-S model validation results revealed that the elevation,slope,distance to the main counties,distance to the main roads,distance to the irrigation canal systems,distance to the drains,multi-year average groundwater depth,gross domestic product,population density,and total agricultural output values had better explanatory on different land use types and were the main driving factors for land use evolution.The simulation accuracy on the 10-year scale and the 30-year scale were both acceptable.The accuracy of the simulation was negative influenced for two main reasons:(i)the simulation involved lots of land use types;and(ii)the farmland salinization had large spatiotemporal variability and complicated driving factors.(4)The results of soil salinization prediction using the CLUE-S model in different scenarios revealed that expanding planting area of wheat,melons,or vegetables would aggravate soil salinization.Moreover,as time went on and the expansion of their planting areas,the accumulation rate of salt in surface soil increased rapidly.Among wheat,melons,and vegetables,the expanded planting area of wheat had the most notable increase of salinization.However,by implementing water-saving irrigation,expanding maize planting area,or expanding sunflower planting area,the average salt content could be efficiently decreased.By conducting 30-year predictions of the three scenarios,the average salt content decreased 7.59%,4.23%,and 3.61%respectively.Thus,the most effective improvement in soil salinization is water-saving irrigation,followed by expanding maize planting area. |
PDF Full Text Request