Study Of Cropland Spatio-temporal Dynamics Change Detection In Hengyang Basin Based On Remote Sensing Time Series Analysis

In recent years,with the continuous acceleration of the urbanization process of Hengyang City and the adjustment of the agricultural planting industry structure,Hengyang’s land cover,especially cultivated land,has undergone tremendous changes.Hengyang City is one of the important commercial grain planting bases in Hunan Province.Therefore,detecting and classifying the land cover changes related to cultivated land in Hengyang City,and describing the long-term temporal and spatial dynamic characteristics of cultivated land,is of great significance for understanding the ecological consequences of cultivated land changes.In this paper,Hengyang City,Hunan Province is selected as the research area.Due to the coupling of traditional change detection algorithms and land cover classification methods,it can only identify changes in land cover types,but fails to identify changes in crop types,making the description of the spatial and temporal dynamics of cultivated land not detailed enough.,Failed to detect changes in the quantity of various crops while considering the amount of cultivated land.The main work and important conclusions of this study are as follows:(1)Using NSPI-based cloud and strip removal methods to interpolate Landsat data,and using a spatio-temporal fusion ESTRAFM algorithm,using MODIS’ vegetation index product to spatio-temporal fusion of missing month data to reconstruct Months are Landsat NDVI time series with time intervals.(2)Use the coupled iterative seasonal trend decomposition model and breakpoint detection time series change detection algorithm-BFAST to carry out change detection on the monthly Landsat NDVI time series to obtain nine seasonal item parameters and two trend item parameters.As input to a random forest classifier,a map of land cover types at any time is generated.The accuracy evaluation results show that the accuracy of change detection reaches 80.2%,the time accuracy evaluation results show that it has high timeliness,and the land cover classification achieves an overall accuracy of 84.17%.(3)Characterize the spatio-temporal dynamics of cultivated land in Hengyang City.The results show that since 2001-2018,the cultivated land area in Hengyang City has not changed much,while the sown area of single-season rice and double-season rice both showed a downward trend.A large area of arable land has been transferred out,and the area close to urban settlements is a hot spot area,indicating that the phenomenon of urban land occupation of arable land has occurred in the process of urbanization.After 2012,Changning City and Leiyang City showed a continuous expansion of cultivated land,and the expansion of cultivated land showed a trend of expanding from a low-altitude plain area to a higher altitude hilly area.The results show that the coupled BFAST algorithm and RFC method can effectively detect land cover changes related to cultivated land and provide information on the type of change,and can provide new research ideas for land cover change detection research involving fine classification of crops.