Spatial-Temporal Variation Of Plantations In Hengyang Based On Landsat Time Series Analysis

Forest is an integrated multi-functional and multi-valuable terrestrial ecosystem,which plays a very important role in maintaining the structure,function and ecological process of the natural ecosystem.With the intensification of global change and the increasing frequency of human activities,the types and distribution of global forests have also changed dramatically.A large number of virgin forests have been cut down,and the planting area of artificial forests has been continuously expanded.As a result,the balance of original ecosystems in many regions has been destroyed and the biodiversity of these regions has been seriously threatened.Therefore,it is of great significance for the protection of the whole terrestrial ecosystem and the sustainable development of human beings to detect the changes of forest cover by remote sensing and extract the temporal and spatial characteristics of plantations.But the current study of plantation is based on high resolution image data or radar data,the data while it is possible to solve the problem of fine classification difficult,but difficult in a wide range of complete data acquisition,data does not have disadvantages such as the duration of continuous time for large regional scale plantation recognition and time series analysis,do not apply.Typical red soil erosion area in China based on a plantation of Hengyang basin as the research object,one of 1988 to 2018 all access to the Landsat remote sensing image as the main research data,spectral time trajectory change detection as the core,with Google Earth Engine cloud platform as the research platform,achieved with the characteristics of the continuous time surviving in high-resolution remote sensing image can be derived from the Landsat distribution range of Hengyang basin plantation,and on the basis of further analysis of the spatial and temporal variation characteristics of plantation,It provides a reference for the dynamic change detection of plantation based on medium resolution remote sensing image.The specific research contents and results of this paper are as follows:Plantation extraction based on Landsat long time series.According to the natural forest and plantation during the period of growth with different vegetation index performance characteristics,selection of NDVI as to distinguish the forest type factor,build time sequence curve,through curve feature selection sample type,mixed characteristics more training classifier,land cover classification using the random forest algorithm extracts the distribution of plantation.In this paper,the difficulty of identifying natural forest and artificial forest pixels in image classification was solved.Based on LandTrendr algorithm,the temporal and spatial variation characteristics of plantations in Hengyang basin were analyzed.Landsat images from 1988 to 2018 were used to construct a data time sequence stack for LandTrendr algorithm to detect forest disturbance changes Hengyang basin during 31 years,and to record the years and durations of disturbance under different disturbance orders.The paper studies the mutual transformation between different ground object types and plantations and analyzes the temporal and spatial variation characteristics of plantations.