Remote Sensing Analysis Of Forest Ecosystem Disturbance And Restoration In Red Soil Erosion Area Using Time Series Data

Forest is an important part of terrestrial ecosystem,which plays a key role in the material cycle and energy flow of the whole biosphere.Due to the influence of soil erosion and other natural factors and human activities for a long time in the red soil region of southern China,the forest ecosystem is constantly undergoing complex changes.It is of great significance to explore the forest ecosystem quality and health in red soil erosion area by excavating the high hidden deep forest disturbance and restoration change information in remote sensing images based on the remote sensing time series.This study is mainly concerned with how to construct a random forest ensemble classification model to divide forest disturbance and restoration into different levels.taking the changes of forest ecosystem in Hengdong County,a typical red soil erosion area in southern China,from 1988 to 2018 as the research object,then constructed the classification model of random forest integrated Land Trendr algorithm and three indexes(NBR,NDVI and TCW),in order to divide the process of disturbance and restoration of disturbed forest into different levels,and the health of forest ecosystem was further evaluated.The main work and conclusions are as follows:(1)Constructe the classification model of random forest integrating Land Trendr algorithm and three indexes(NBR,NDVI and TCW).The process of disturbance and restoration of disturbed forest was divided into different levels.The magnitude of disturbance was divided into high,moderate and light,and the magnitude of recover was divided into complete recover,partial recover and no recover.The spatial distribution of different levels of forest disturbance and restoration was then obtained.The results show that the ensemble classification model of random forest has a good classification effect,and the classification accuracy of different levels of forest disturbance and restoration reaches 80.36% and 81.18% respectively.(2)Analyze the results of different forest disturbance and recover levels.The nine forest change patterns in the study area showed fragmentation distribution,and the main change patterns were LD-FR and MD-FR.MD-NR and HD-NR accounted for the smallest proportion of area.Furthermore,the relationship between different forest change patterns with meteorology and topography was discussed.It was found that temperature has no great influence on the intensity of forest disturbance,while precipitation and terrain factors have a certain influence on the process of forest disturbance.For the restoration process of disturbed forest,the change intensity level will be affected by temperature,precipitation and topography.(3)Evaluation of forest ecosystem health level.Based on the classification results of different change patterns of forest disturbance and restoration,the health level of forest ecosystem in Hengdong County was qualitatively evaluated.The forest health level is divided into five levels which are high quality,good,healthy,sub-health and unhealthy.The results show that 85.67% of Hengdong County's forest health level was in the healthy and above level,and the area of sub-health and unhealthy was less.Therefore,the health level of forest ecosystem in Hengdong County is at a higher level,which has the typical characteristics of healthy forest ecosystem such as sustainable stability and strong self-healing ability.It is the result of long-term mutual adaptation between ecological environment and various species communities in forest.