| Land use/ Land cover change(LUCC)is one of the main causes of global environmental change,and is also the focus and hot spot of global change research.In recent years,due to the increasing population and unreasonable land use practices,the desertification of the Horqin Sandy Land has further intensified and the ecological environment has become more fragile.In order to guide the rational use of local land resources and promote the ecological sustainable development of the region,this study used Landsat series satellite images to obtain land cover classification maps for2000,2010 and 2019 based on machine learning classification methods,and explored the spatial and temporal land cover variation characteristics and its driving factors in the horqin sands from 2000 to 2019.Firstly,this study preprocessed the selected remote sensing images with image cropping and stitching.Then,the land cover types in the study area were classified into seven categories: cropland,grassland,woodland,sandy land,water area,bare soil and built-up land.The sample points of the corresponding categories were selected based on the field survey images and other auxiliary image data,and randomly divided into training and validation sample points.Based on the selected sample points,the Random Forest method(RF)and Support Vector Machine method(SVM)were used to classify the land cover,and the classification result maps were obtained in two phases.Then,the overall accuracy(OA),user’s accuracy(UA)and producer’s accuracy(PA)of the RF method and SVM method were calculated separately by establishing confusion matrices,so as to evaluate and compare the classification results of the two methods.In order to better explore the land cover change characteristics of the Horqin Sandy Land,we used the grid cell method to quantitatively analyze the relationship between each land cover type by establishing 900 m×900 m grid cells after obtaining the land cover result maps of the two phases.Finally,this research also further explores the driving forces of land cover change in the region by combining natural factors such as temperature and precipitation and social factors such as population,industrial economy,technology and national policies.The results show that:(1)Comparing two land cover classification methods,Support Vector Machine(SVM)and Random Forest(RF),the land cover classification results in 2000 and2019 showed that the classification accuracy of SVM was generally higher than that of RF(the overall accuracy of SVM method and RF method in 2000 was 88% and85.9%,respectively,and the overall accuracy of SVM method and RF method in 2019 was 87.9% and 80.0%).Therefore,the land cover classification maps obtained by the SVM classification method for 2000,2010 and 2019 were used for subsequent analysis in this paper.Among them,the overall accuracy of the land cover classification for 2010 was 80.55%.(2)After quantitative analysis of land cover change,from 2000 to 2010,the area of cropland,grassland and built-up land in the Horqin Sandy Land increased,with the most significant increase in the area of cropland and the decrease in the area of woodland,sandy land,bare soil and water area,with the most significant decrease in the area of water area.From the analysis of land transfer,the period mainly shows the conversion of grassland,woodland and bare soil to cropland,and the conversion of bare soil and woodland to grassland.From the results of correlation analysis,the change of cropland in the main area of sandy land showed a high negative correlation with the change of bare soil area(r=-0.32),and the correlation of the change of cropland area with the change of grassland and woodland area was weak.This indicates that although some grassland and woodland were reclaimed as cropland during the period,the expansion of cropland in the main area of the Horqin Sandy Land from 2000 to 2010 was mainly based on the reclamation of bare soil.The change of grassland area has a strong negative correlation with the change of bare soil area(r=-0.61)and the change of sand area(r=-0.39)indicates that the ecological restoration and desertification control of the horqin sands have made some progress during this period.(3)From 2010 to 2019,the cropland in the Horqin Sandy Land increased and the grassland,water area,woodland,sandy land,bare soil and built-up land decreased,with the most significant decrease in water area.In terms of land transfer,it shows the conversion of grassland,bare land and waters to cropland and grassland to bare soil.From the results of correlation analysis,there is a high negative correlation between the change in cropland area and the change in bare soil area(r=-0.36),and a weak correlation between the change in cropland and the change in grassland(r=-0.19)and waters area(r=-0.06).This indicates that the expansion of cropland in the period was still dominated by the reclamation of bare soil.The change in grassland area has a strong negative correlation with the change in bare land area(r=-0.69)and also with the change in sandy land(r=-0.12),indicating that ecological restoration in the area still needs attention after 2010.(4)Although natural factors such as temperature and precipitation play a role in the reduction of bare sand and water in the Horqin Sandy Land,they are not the main driving force.Population,industrial economy,technology and policy are the main drivers of land cover change in this area over the past 20 years.a)From the factor of demographic,population growth has led to an increasing demand for food,and local farmers and herdsmen have reclaimed large amounts of grassland,woodland and bare soil into cropland,resulting in a rapid increase in cropland.b)From the factor of industrial economy,economic development and industrial structure adjustment and upgrading have made the output value of the secondary industry exceed that of the primary industry and become the dominant industry in the Horqin Sandy Land,making more bare soil to be developed into built-up land.Meanwhile the area of grassland,woodland and bare soil have reduced.c)From the factor of technology,the increase in the level of agricultural mechanization and the development of agricultural irrigation technology have also contributed to the expansion of cropland in the region.With the increase of water consumption and water usage in the upper reaches of the rivers,etc.,the water area in the region has also been significantly reduced and the rivers have dried up,providing the material basis and development space for the development of cropland.d)From the factor of policy,in the early 1980 s,the agricultural and animal husbandry industry in the Horqin Sandy Land implemented the production responsibility system,such as "contracting production to households" and "contracting farmland and grassland",which led to a rapid increase of cropland.And after 2000,the ecological restoration projects such as returning farmland to forests and grasses were implemented in the area.As a result,a large amount of bare soil was converted to grassland from 2000 to 2010,the area of grassland increased during this period,and the ecological environment of the horqin sandy area developed toward a benign one.In addition,due to the internal imbalance of the ecosystem,the government implemented an ecological migration policy in this area,concentrating herders who were originally located in areas with severe ecological damage after transferring them,and this strategy led to a tendency to reduce the built-up land from2010 to 2019.Based on a comprehensive grasp of the relationship between various driving forces and land cover change,this study will deepen the interrelationship between land,population,policy and development in the Horqin Sandy Land,and has some reference significance for the rational use of land resources in this region. |