| Cultivated land is an important foundation of agricultural production and has a significant impact on national food security and long-term stability.Remote sensing technique,as its name suggests,can remotely sense both spatial distribution and changes of large cultivated land in a very timely and accurate manner.It is of great significance to advance agricultural modernization and informationization as well as agricultural planning and policy formulation.In order to solve the problem of slow land coverage acquisition and low efficiency of traditional cultivated land over a long time,this thesis utilized Google Earth Engine(GEE)cloud platform and Landsat remote sensing images as research data to study cultivated land coverage issue in Jiangsu Province over a period of 20 years from 2001 to 2021.Disturbance and spatial analyses were carried out to study the range,intensity and transfer of cultivated lands in Jiangsu Province in the past 20 years.Combined with the Jiangsu Provincial Statistical Yearbook,a model was established to analyze the driving force of cultivated land change.The specific research contents and major conclusions derived are:(1)Based on the GEE cloud platform,multi-dimensional feature data set integrating spectral index,texture information and phenological characteristics was established,and a high-precision land classification model was developed according to the geographical distribution characteristics of Jiangsu Province.Experiments were designed to extract cultivated land in Jiangsu Province by using four kinds of classification methods optimized by tuning parameters,namely support vector machine.The accuracy test shows that the random forest algorithm has the highest accuracy,the overall accuracy is 95.79%,the Kappa coefficient is 0.94,and the error and missing error are 3.38% and 8.86% respectively.Finally,the random forest algorithm was used to extract the land use type and cultivated land range in the area from 2001 to 2021.(2)Based on the LandTrendr algorithm and Geographic Information System(GIS)spatial analysis method,the spatial and temporal changes of cultivated lands were analyzed.The LandTrendr algorithm was used to detect the interannual land use change,and the LandTrendr time series map was drawn to investigate the intensity of cultivated land disturbance in Jiangsu Province.The land use transfer matrix of the study area(in5-year time span)was studied by using GIS spatial analysis,and the inflow and outflow of cultivated lands in Jiangsu Province was analyzed.It was found that urbanization construction was the main reason for the outflow of cultivated land in the past 20 years.In the early period,the proportion of impervious surface of cultivated land was high in southern Jiangsu,and after 2011,there was a large outflow of cultivated land in central and northern Jiangsu.The main source of cultivated land inflow is water body and forest land development.And the single dynamic attitude of cultivated land was-0.66% and-0.25%,respectively,which showed that the cultivated land area of Jiangsu Province showed a downward trend and the dynamic attitude of cultivated land decreased.(3)The driving force analysis model of long-term cultivated land changes in Jiangsu Province was established.The cultivated land area changes and related driving factors in the study area were statistically analyzed.Principal component analysis was used to obtain two principal components of the driving factors,and the cumulative contribution rate reached 87.55%.The comprehensive score of the driving force and the weight of each driving factor were calculated.A multiple linear regression model of cultivated land changes was established to investigate the influence mechanism of each driving factor on the changes of cultivated land area in the study area.Based on the variable coefficient of the regression equation,the cultivated land area of Jiangsu Province decreases with the increase of population growth rate,grain output and average annual precipitation,and it is expected that the cultivated land area of Jiangsu Province will increase with the increase of the proportion of the secondary industry and the increase of machine-cultivated land. |