Remote Sensing Simulation Of Evolutionary Mechanism Of Phyllostachys Edulis Expanding Chinese Fir Forest

Phyllostachys pubescens and Cunninghamia lanceolata,as two widely distributed tree species in the south,often grow adjacent due to their similar growth environment and site conditions.Phyllostachys pubescens have the advantage of growing normally in adjacent closed forests due to their underground stems that can provide the necessary nutrients for their own growth through the mother bamboo whip root system and root system,enabling them to expand to other stands.Due to the inability of remote sensing technology to detect the growth of underground stems,the expansion process of Phyllostachys pubescens can cause adaptive changes in the aboveground forest.By analyzing the forest indicators that can characterize the expansion changes of Phyllostachys pubescens,the quantifiable characteristics of Phyllostachys pubescens expansion evolution through remote sensing are determined,and a prediction model for Phyllostachys pubescens expansion is constructed to lay the foundation for remote sensing monitoring of Phyllostachys pubescens expansion.Therefore,the study takes the mixed forest of Phyllostachys pubescens and Cunninghamia lanceolata in Tianbaoyan Nature Reserve,Yong’an City,Fujian Province as the research object.Based on the proportion of bamboo from 0-20%,20-40%,40-60%,60-80%,and 80-100%,it is divided into five expansion degrees.Using visible light and multispectral image data obtained by drones,combined with ground measurement data,remote sensing inversion and differential quantification analysis of evolution characteristics are achieved,and a prediction model for bamboo expansion is constructed,In order to achieve remote sensing monitoring of bamboo expansion process and provide reference for bamboo expansion prevention and control.The main conclusions are as follows:（1）During the process of expanding the Chinese fir forest with bamboo,bamboo shoots grow through the expansion of bamboo whips,causing changes in the canopy height of the forest.Through analysis of the differences in canopy height between the years 2020 to 2021 and 2021 to2022,it was found that using the height differences caused by the emergence of new bamboo can achieve remote sensing quantification of changes in the canopy height of the forest.The results showed that the overall accuracy of remote sensing quantification in study area 1-4 was80.0%,84.4%,78.3%and 76.0%,respectively.（2）Through correlation analysis,eight vegetation indexes with high correlation with leaf area index（LAI）,namely DSI_（1,5）,DSI_（5,2）,DSI_（5,3）,DSI_（5,4）,EVI,MSAVI,MTVI and OSAVI,were selected to participate in the modeling.Multiple linear regression,principal component analysis,machine learning（random forest,support vector machine,BP neural network）and curve estimation based on single vegetation index were used to build the estimation model of stand LAI,and R²,RA,RMSE to determine the best estimation model.The results show that the LAI estimation model based on random forest has the best effect,with R²=0.583,RA=88.8%,RMSE=0.342.Therefore,the random forest model is used for inversion of LAI in the study area.（3）Five vegetation indexes(RSI_（1,5）,RSI_（2,4）,OSAVI,RSI_（1,4）,and RSI_（3,4）),which are highly correlated with soil fertility,were selected to participate in the construction of remote sensing estimation model of soil fertility.Random forest（RF）,Support Vector Machine（SVM）,and BP neural network were used to construct soil fertility estimation models.The results indicate that the BP neural network has the best estimation performance,with its R²=0.421,RA=78.7%,RMSE=0.108,so the research selects a model based on BP neural network to realize remote sensing inversion of forest soil fertility.（4）Based on images fused with multiple spectral features in the visible light,red edge,and near-infrared bands,the local maximum method is used to extract tree vertices in the study area,and combined with the canopy height model（CHM）,the position and height of individual trees are obtained.Based on the combination of four image features,namely,visible light+spectral feature,visible light+texture feature,visible light+CHM and visible light+spectral feature+texture feature+CHM,tree species recognition and classification are realized through object-oriented multi-scale segmentation,combined with three machine learning classification methods,namely,K neighborhood method（KNN）,random forest,and support vector machine.Using the obtained individual tree positions,construct Voronoi diagrams and Delaunay triangulation networks to obtain forest spatial structure units.Select five spatial structure indices,including size ratio,degree of mixing,angular scale,openness ratio,and competition index,to construct a forest spatial structure evaluation index and achieve remote sensing inversion of forest spatial structure.The results showed that the overall accuracy,misclassification error,and missed classification error of single tree extraction in study area1-4 were 88.5%,11.5%,9.7%,83.2%,16.8%,15.3%,87.0%,13.0%,7.8%,and 85.8%,14.2%,11.4%,respectively;Under the combination of visible light,spectrum,texture,and CHM image features,the SVM classification method performs well in tree species recognition and extraction,with an overall accuracy and Kappa coefficient of 87.0%and 0.72,respectively.（5）Based on the evolution feature indicators obtained from remote sensing,a partial least squares regression was used to construct a prediction model for bamboo expansion.The model has R²=0.582,RMSE=0.018,and RA=74.16%.With the expansion of Phyllostachys pubescens to Cunninghamia lanceolata forest,the changes of stand leaf area index,stand canopy height,stand soil fertility,and stand spatial structure are different to some extent:the changes of stand leaf area index and stand canopy height show an increasing trend,the soil fertility of the stand shows a process of first decreasing and then increasing,and the stand spatial structure shows a trend of first increasing and then decreasing.By comparing and analyzing the sample plots with changes in expansion degree from 2021 to 2022,it was found that changes in forest canopy height and forest leaf area index have a positive impact on expansion degree;The spatial structure and soil fertility of forest stand are negatively affected.The changes in forest leaf area index and forest canopy height play a major role in the expansion and evolution of Phyllostachys pubescens.