Monitoring Cropping Intensity Dynamics Of Cultivated Land And Its Driving Factors In Northern China

As a new space science technology,remote sense can be widely used in regional agricultural condition monitoring because of its characteristics of high timeliness,low cost and wide range.By analyzing the changes of cultivated land and its CI information in the single cropping and double cropping system transition region across northern China(mainly located in Beijing,Tianjin,Gansu Province,Shaanxi Province,Shanxi Province and Hebei Province),the laws and characteristics of temporal and spatial differentiation of farming system in this area are obtained,and the main influencing factors of its dynamic change are explored.These urgent works are conducive to the rational utilization of barren cultivated land resources,the scientific formulation of agricultural development decisions and the effective protection of sensitive ecological environment in the agricultural pastoral ecotone of northern China.Meanwhile,these are also of great significance to ensure the realization of sustainable development goals(SDGs)"eliminating hunger,realizing food security,improving nutrition and promoting sustainable agriculture".Taking single cropping and double cropping system transition region in northern China as the research object,this study carried out long time series remote sensing monitoring of cultivated land and its CI information and then explored the driving factors which influenced the spatial differentiation characteristics of northern boundary in double cropping system(DCS)across this area.The main work includes the following aspects.Using the advantages of Google Earth engine(GEE)cloud platform in remote sensing data processing and analysis,the long-time series cultivated land information of the study area from 2000 to 2020(totaling 11 periods)was obtained by the construction of recognition feature set to cultivated land based on multi-source remote sensing satellite images,and the utilization of random forest classifier.Then,based on the long-time series data of cultivated land and MODIS NDVI from 2000 to 2020 in the study area,the cultivated land CI information of 11 periods was extracted used the quadratic difference method.Finally,based on the spatial distribution data of CI,this paper obtained DCS range and the spatial-temporal distribution data of its northern boundary by utilizing nuclear density estimation method and Densi-Graph method,and on this basis,the influence intensity of human activity factors and natural elements to the northern boundary of DCS by utilizing Pearson correlation analysis and Geo detector model.The conclusions of the paper mainly include the following aspects.(1)The study completed a method for fine-grained mapping of cultivated land information in the study area for 2000-2020 based on an improved random forest classifier.Based on GEE cloud platform,it can realize the rapid and reliable recognition and spatiotemporal feature analysis of cultivated land information with large-scale and long-term 30m spatial resolution in the study area by the use of random forest classifier.The overall accuracy of cultivated land remote sensing identification products obtained from GEE is higher than 90%and the Kappa coefficient is more than 0.8 based on the confusion matrix.Compared with the cultivated land area of the statistical data,the overall accuracy is mostly between 65～80%,indicating that the results of this study are reliable.In terms of spatial distribution,the cultivated land resources in the study area had obvious dominant geomorphic characteristics.They were mainly distributed in the plain and hilly areas with an altitude of less than 3500m,a slope of less than 15°and a topographic potential index of less than 1.24.Its distribution center was located in Lishi County,Shanxi Province,distributing between 111°25′25″–111°36′8″E、37°20′29″–37°26′16″N,and it had migrated 12.88km to the northwest at the speed of 613.33m/a from 2000 to 2020.In terms of time change,the total area of cultivated land resources in the study area showed a small reduction trend from 2000 to 2020.The largest area of cultivated land was in 2000,which was 3,014.09×10~4hm~2,and it decreased by 188.17×10~4hm~2 in 2020.The dynamic degree of cultivated land use from 2000 to 2020 was-0.31%,showing a small range and less trend.Excepting Shaanxi Province,the cultivated land area of all provinces and cities showed a decreasing trend.(2)The study realized the fusion of MODIS vegetation index and Landsat/Sentinel-2 cultivated land information based on MCI as a measure for long time series remote sensing monitoring of CI.Based on the long-time series dynamic change information of cultivated land in 30m spatial resolution,it can improve the remote sensing spatial mapping accuracy of MCI to a certain extent,and obtain more accurate CI information and its temporal-spatial change characteristics from 2000 to2020 in the study area.The results show that the MCI obtained in this study is in good agreement with the MCI calculated by statistical data,with about 87.88%of the MCI remote sensing inversion accuracy exceeding 80%.In terms of spatial distribution,the MCI of cultivated land in the study area showed an overall trend of high in the southeast and low in the northwest,with the highest mean MCI of 127.93%in Hebei Province and the lowest MCI value of 108.74%in Gansu Province.In terms of temporal changes,the proportion of arable land in the study area with MCI of 100%was significantly larger than that with MCI of 200%from 2000 to2020,and the area with MCI of 200%showed a fluctuating increase,with an average annual increase of about 0.4%(r=0.0.82,P<0.001).The MCI in Beijing and Shaanxi Province showed a decreasing trend in the proportion of DCS,while it showed an increasing trend in Tianjin,Hebei,Shanxi and Gansu provinces.(3)Based on the spatial distribution of MCI,the study detected sensitive geographic areas for the change of the northern boundary of DCS,and analyzed the main drivers of the spatial and temporal evolution of the northern boundary based on geodetector factor detector model and Pearson correlation coefficients.The northern boundary of DCS mainly changes between 33°34′20″N–40°40′40″N,102°42′28″E–119°48′14″E,showing the dynamic characteristics of moving south first and then moving north(the degree and distance of moving south are larger than that of moving north).And the spatial differentiation pattern and temporal-spatial change characteristics of DCS northern boundary are jointly affected by natural factors and human activities.From the Pearson correlation analysis,it can be seen that the average March temperature and potential evapotranspiration have the greatest influence on the distribution range of the northern boundary of DCS among the drivers of natural elements,with Pearson correlation coefficients(r)of-0.692 and-0.622,respectively.And among the human activity elements,the total power of agricultural machinery has the greatest influence with r of 0.387.From the geodetector factor detector model,it can be seen that the landform type,slope,topographic position index,average July temperature and average growing season temperature have a strong impact on the spatial differentiation pattern of the northern boundary of DCS in the study area.The maximum q value is 0.173 and the lowest q value is 0.001.The combination of landscape type and agricultural fertilizer application has the greatest explanatory power of 22.3%,and the influence of distance to road,distance to water system,and population density with other factors is relatively low.The interaction between distance to road,distance to water system,population density and other factors was relatively low,with the weakest explanatory power of 1.1%.