The Forest Structural Parameters Estimation And Growth Process Simulation Of Ginkgo Plantation

Planted forests is significant constituent part of forest ecological system,which plays a vital role in supplying national forest timber products and providing ecological benefits such as maintain carbon balance,mitigate the green house effect.China has the largest area of planted forests all over the world,forest structural parameters are the most basic quantitative characteristic of forest plantation ecosystem,planted forest structural attributes can reflect tree species characterstics,site quality,and management level of forest plantation;the study on the growth process of planted forest offered the information of the dynamic change of forest stand growth,the response of plantation growth to cultivation measures of forst managers,and the prediction of potential productivity of plantation.Therefore,accurate and reliable acquisition of planted forest structural attributes and the information of growth process are significantly important for forest managers to make decisions on long-term sustainable forest management.Traditionally,the measurement of forest structural parameters were time consuming and costly,and usually only a limited number of samples can be acquired and implemented,based on the remote sensing technology of UAV-LiDAR,stand-level forest structural parameters could be accurately estimated due to its capacity to provide highly accurate estimations of three-dimensional（3D）forest structural information.Applying process-based model and climate model,the growth rhythm could be accurately described,the dynamic change of forest stand growth could be predicted,which offered a new technique and method for precise cultivation of planted forests,and provide data support for the efficient use of forest plantation.Ginkgo（Ginkgo biloba L.）tree is a well-known economic tree species and is one of the dominant species in Chinese timber forest,accurate and reliable acquisition of ginkgo forest structural parameters and the information of growth process are significantly important for precise cultivation and efficient use of ginkgo plantations.In this study,Different growth processes of ginkgo trees were fitted using a stem analysis method.At the individual tree level,we focused on biomass allocation patterns of ginkgo trees and the allometric models for each component biomass were developed;UAV-LiDAR data was used to estimate forest parameters in ginkgo plantation,besides,a simple process-based model of 3-PG was applied to predict growth of ginkgo plantaion in northern Jiangsu Province.The main results are as follows:（1）The DBH of the 27-year-old ginkgo tree in Pizhou was 25 cm,and the current annual growth reached a maximum（1.15 cm）at the age of 12.In Dongtai,the DBH of the 20-year-old ginkgo tree was 12.4 cm,and the current annual growth reached a maximum（0.79 cm）at the age of 9.According to the intersection of the average volume increment and current annual volume increment,the thinning age in these two areas should be 20 and 17 years,respectively.At the individual tree scale,the wood density had a significant positive correlation with the ring position（R²=0.93,p<0.01）and relative growth rate（R²=0.71,p<0.01）,and at the stand scale,the basal wood density had a significant positive correlation with the DBH（R²=0.86,p<0.01）and tree height（R²=0.60,p<0.01）.In plain areas,the Gompertz model provided the best prediction of DBH（R²=0.998,SSE=1.55 cm）,height（R²=0.998,SSE=1.11 m）,and stem biomass（R²=0.998,SSE=3.37 kg）,while the logistic model provided the best prediction of volume（R²=0.972,SSE=2.15 m³）.In coastal areas,the Gompertz model had the best performance in terms of predicting DBH（R²=0.998,SSE=1.12 cm）and height（R²=0.998,SSE=0.94 m）,compared with the Korf model for volume（R²=0.996,SSE=1.98 m³）and stem biomass（R²=0.998,SSE=1.91 kg）.In addition,we found that there are no remarkable difference of bulk density between these two regions,whereas the content of soil organic matter（SOC）,total N,total P,and total K in Pizhou forests was higher than them in Dongtai forests.（2）The whole plant biomass variation range of the ginkgo trees（with DBH ranging from 10cm to 27 cm）was 28.50-320.27 kg for each tree,respectively.Relative proportions of stem,branch,leaf,and root to total tree biomass were 49.4%—56.6%,12.1%—18.9%,3.8%—5.5%,and 26%.The aboveground biomass was significantly linearly correlated with belowground biomass.The slope of the fitted linear model was 0.35.Results showed that the majority leaf and branch biomass occurred in the middle canopy layers,with significant difference between the middle and upper and lower layers in combined biomass of leaves and branches,and there is no significance between upper and lower layers.For all sample trees,about 70%of root biomass were observed in the 0—1.0 m soil layer,with soil depth increasing the root biomass decreased exponentially.At branch level,allometric models based on two variables（i.e.BD and BL）of branch biomass explained more than 95%of the variations in data.The result showed that D was a best independent variable in estimating the biomass of leaf,branch,aboveground section than the rest variables,and D—H was the best in estimating stem,root and total tree biomass.（3）In general,models based on both plot-level and individual-tree-summarized metrics（CV-R²=0.66-0.97,rRMSE=2.83%-23.35%）performed better than models based on the plot-level metrics only（CV-R²=0.62-0.97,rRMSE=3.81%-27.64%）.PLS had a relatively high prediction accuracy for forest paramters（except Lorey’s mean height）than MLR,MLR had a highest prediction accuracy for Lorey’s mean height（CV-R²=0.97,rRMSE=2.83%）among these four approaches,whereas k-NN performed well for predicting volume（CV-R²=0.94,rRMSE=8.95%）and AGB（CV-R²=0.95,rRMSE=8.81%）.For the point cloud density sensitivity analysis,the highe-related metrics showed a little correlation with point cloud density,while the canopy volume metrics showed a higher dependence on point cloud density than other metrics.The correlations between AGB and the metrics of height percentiles,lower height level of canopy return densities and canopy cover appeared stable across different point cloud densities when the point cloud density was reduced from 50%(80 pts·m^-2)to 5%(8 pts·m^-2).ITD results showed a relatively high accuracy（F₁-score>74.93%）when the point cloud density was higher than 10%(16 pts·m^-2).（4）A simple process-based model of 3-PG（Physiological Principle Predicting Growth）model based on physiological principle and environmental factors was calibrated and applied to simulate DBH,stem density,volume and AGB growth over a ginkgo plantation in northern Jisangsu Province.The results showed that:according to the growth prediction of 3-PG model,the DBH was 47 cm,stem density was 250 N/ha,volume was 135 m³/ha,and AGB was 90 mg/ha in the first50 years.3-PG had a quite good performance for simulating forest structural attributes with the R²>0.90,rRMSE<10%except for stem density（R²=0.84,rRMSE=19.03%）.Based on the validation data,the prediction values simulated from 3-PG is higher than the observed values of forest parameters,with the validation accuracy of R²>0.8,rRMSE<16%for DBH,volume and AGB.For the sensitivity analysis of fullCanAge and FR,they showed that these two parameters are vital for 3-PG model,and they have a positive correlation with the simulated results.