Dynamic Monitoring Of Croping System In Anhui Province Based On Time-series Remote Sensing Images

The change of planting system is affected by social economy,climate change and other factors,so the study of its regional distribution and spatial-temporal evolution characteristics will play a positive role in the sustainable development of agriculture and national food security.Under complex cropping structures and fragmented tillage,high resolution mapping of cropping system is an important means to reduce uncertainty.This paper took Anhui Province as the research area,dryland,drought-flood rotation,singlecropping rice and double-cropping rice as the research objects,based on the cloud platform of Google Earth Engine and time-series Landsat and Sentinel remote sensing data,field investigation and visual discrimination methods were used to collect samples.In order to remove image noise and ensure information richness,A method combining30-day synthesis with 15-day step length,gap filling and S-G filtering was proposed to reconstruct the spectral index of time series,and the recognition accuracy of random forest,support vector machine and gradient lifting regression tree model under various band combinations was compared and evaluated.The optimal model was selected to draw30 m resolution cropping system maps of Anhui Province in 1990,2000,2010 and 2020.The model recognition results based on NDVI+MNDWI showed good accuracy,with the optimal overall accuracy reaching 85%,90.6%,90.1% and 93.4%,respectively.Compared with the existing land use type data,there’s been a huge increase in precision and accuracy.In order to analyze the temporal and spatial characteristics of crop/cover change in Anhui Province,based on the identification results of cover types from 1990 to 2020,land use transfer matrix and land use dynamic attitude were used to analyze the characteristics of land use transfer in Anhui Province in recent 30 years.Based on the method of gravity center analysis,standard deviation ellipse and land use landscape pattern,the spatial pattern change of land cover types in Anhui province was revealed.Combined with getisOrd General G clustering method and multi-distance spatial clustering(Ripley’s K function)method,the temporal and spatial patterns of land use change accumulation in Anhui Province were analyzed.The results showed that:(1)from 1990 to 2020,the cultivated land area in Anhui province decreased by 3.8%,and the changes of dry land,drought-flood rotation,single-cropping rice and double-cropping rice were-0.68%,-7.66%,0.26% and-51.40%,respectively.The phenomenon of "double change to single cropping" was obvious.There were frequent changes between drought-flood rotation and single-cropping rice,and the impervious water surface and bare land increased by 55%.(2)The distribution characteristics of dryland were stable,and the distribution range of paddy and upland rotation was similar to that of single-cropping rice,but the landscape pattern of double-cropping rice changed greatly in different periods.The overall landscape pattern tended to be fragmented,scattered,diversified and complicated,and the spatial distribution and area tended to be homogenized.(3)The drastic change of land use has a significant aggregation,and different degrees of change can be better classified by Kmeans clustering.The aggregation characteristics of the region with severe land use change are less affected by the observed scale change than that of the region with gentle change.Figure [30] Table [8] Reference [104]...