Growth Prediction Of Chinese Fir Plantation Based On Random Forest Algorithm Under Climate Change

Under the background of climate change,forests,as an important part of terrestrial ecosystems,play a huge role in mitigating the impact of global warming.Forest growth and harvest model is an important tool for forest management decisions,which can provide suitable management strategies for forest managers to cope with climate change.Random forest（RF）algorithm,which has the advantages of no assumptions and can deal with complex variables,is widely used in research.The application of random forest algorithm to construct the growth and harvest model of climate-sensitive Chinese fir（Cunninghamia lanceolata）plantation has important theoretical significance for correctly evaluating the source and sink function of Chinese fir plantation,and establishing the sustainable management of Chinese fir plantation and the adaptive management system to cope with climate change.In this paper,the Chinese fir plantation in Dabie Mountain area and southern Anhui area was taken as the research object.The typical sample plot survey method was used to obtain the sample plot data.The Wolrd Clim2 was used to extract the climatic factor data,and the Soil Grids 250 m system was used to obtain the soil data.The random forest algorithm was used to construct the climate-sensitive stand growth model system(stand average DBH,average tree height,plant density,sectional area,accumulation,biomass（including aboveground,underground and total biomass）.Through the climate sensitive growth model system,the potential growth of Chinese fir plantation in the study area was predicted under three future climate scenarios（ssp126,ssp245,ssp585）in the near and medium term（2021-2080）.The main research results of this paper are as follows:（1）Construction of climate sensitive stand average DBH growth model.The corresponding model of the optimal super-parameter combination was shown as R²_cv=0.8911,RMSE_CV=0.02m.This model can be used to predict the average DBH of Chinese fir plantation.（2）A climate-sensitive stand average tree height growth model was constructed.The corresponding model of the optimal super-parameter combination was shown as R²_cv=0.8742,RMSE_CV=1.3452m.This model can be used to predict the average tree height of Chinese fir plantation.（3）A climate-sensitive stand tree density growth model was constructed.The corresponding model of the optimal super-parameter combination was shown as R²_cv=0.6333,RMSE_CV=390.3542trees.hm^–2.It is proved that the stand tree density model responded to climate factors.（4）A climate-sensitive stand area growth model was constructed.After the density of climate-sensitive stand trees was brought into the model construction,the corresponding model of the optimal super-parameter combination was shown as R²_cv=0.758,RMSE_CV=5.7317m².hm^–2.It is proved that the stand area model responded to climate factors.（5）A climate sensitive stand volume growth model was constructed.After the climate-sensitive stand density and stand section area were brought into the model construction,the corresponding model of the optimal super-parameter combination was shown as R²_cv=0.9317,RMSE_CV=30.1471m³.hm^-2.It is proved that the accumulation model has a response to climate factors.This model can be used to predict the cumulative growth of Chinese fir plantation.（6）After the climate-sensitive stand tree density,stand section area and stand volume were brought into the model construction,the optimal hyper-parameter combination corresponding model of aboveground,underground and total biomass was obtained as R²_cvranged from 0.9465 to 0.981,RMSE_CV ranged from 2.2527 to 9.1269t.hm^-2,which proved that the stand biomass model responds to climatic factors.This model can be used to predict the biomass growth of Chinese fir plantation.（7）By constructing a climate-sensitive stand model system,the stand growth under different future climate scenarios from 2021 to 2080 was simulated and predicted.Average DBH,average tree height,volume and biomass（including aboveground,underground and total biomass）showed positive correlation growth under three future climate scenarios.Overall,the future growth simulation of average DBH and average tree height was ssp126>ssp585>ssp245;the future growth simulation of accumulation was ssp245>ssp126>ssp585;the future growth simulation of aboveground,underground and total biomass was ssp126>ssp245>ssp585.