Forest Aboveground Biomass Estimation And Evaluation Using Multiscale Data Across The Miyun Reservoir Basin In Beijing,china

Forest aboveground biomass is the most important terrestrial carbon pool which reserved over 80% of the Earth surface's carbon stock.It is also an important indicator for assessing the healthy development of forest ecosystem and the level of its net primary productivity.Given the complex development circumstance characterized with global climate change and regional environmental pollution,global studies on different forest ecosystems have suggested that the accumulation of aboveground biomass are experiencing varying degrees of influences,with either increased or decreased overall biomass levels,and increasing or decreasing accumulation rates.Miyun Reservoir Basin,which is selected as the study area in the current thesis,is located in the outskirt of Beijing.Its forest development might has been substantially affected both by the climate change and by the rapid urban expansion and the resulting pollutions.Thus,it is vital to unpack the distribution of forest aboveground biomass and its temporal variations in the area for understanding the response of local forest ecosystem to the climate and environmental changes.Different methods have been developed for forest aboveground biomass estimations.These includes calculation based on forest inventory data,calculation based on logistical models and calculation based on remotely sensed images.In this thesis,the history of the methods,as well as their pros and cons and applications were comprehensively reviewed.Then,the remote sensing based method were selected to quantify the biomass within the Miyun Reservoir Basin,with the assistance of plot measured metrics and tree ring series.In specific,the thesis has been carried out according to the following workflow:(1)Plot sampling.61 plots were set and diameters at breast height,tree height were measured for each single tree within the plots.Species specific allometric equations were adopted to quantify tree-level biomasses.Plot-level aboveground biomass were calculated by summing up the tree level results.(2)Aboveground biomass estimation with LiDAR data.Li DAR point cloud data were processed to calculate the canopy structural metrics including quantile height,point densities etc.The metrics were related to plot collected biomass using stepwise regression analysis.The established regression models were applied to estimate aboveground biomass within the LiDAR flight areas,which were subsequently used as ground truth data to train the prediction models for regional biomass estimations.(3)Regional aboveground biomass estimation with Landsat.The relationship between measured biomass and Landsat spectral indices were quantified through stepwise regression analysis.Forest aboveground biomass levels at three distinct periods(1990,2000 and 2010)were determined with the established regression model.Spatiotemporal characteristics of the biomass were interpreted.(4)Accuracy evaluation with tree ring documents.In this part,the quantitative relationship between tree age,diameters at breast height and tree height were determined for reconstructing the diameter and height series of the trees in the sampling plots.Plot scale aboveground biomass were further calculated using the same set of allometric equations.The biomass results in 1990,2000 and 2010 were extracted and related to Landsat based estimations for evaluating the accuracy of remote sensing based method.Through carrying out the proposed experiments on the forest ecosystem in the Miyun Reservoir Basin,this thesis found that LiDAR point cloud data could obtain canopy structure parameters with high accuracy which subsequently created relative precise aboveground biomass estimations(R2 = 0.85,RMSE = 0.2216).With sufficient training data and evaluation data provided by LiDAR derived estimations and plot measurements,the thesis established a robust aboveground biomass estimation model for the forest ecosystem in the studied area.The model outputs indicated that the average biomass density of the area was about 59.51 t / ha and the total aboveground biomass was valued at 4.5 × 107 t.From 1990 to 2010,the forest ecosystem in Miyun Reservoir Basin showed an overall increase,as evidenced from increasing total forest area,rising average biomass density and elevated biomass level.The spatial and temporal variations of aboveground biomass during the three periods also reflected the possible contributors.In general,the overall development of the forest ecosystem could be attributed to the implementation of several ecological engineering measures,while the terrain variations,along with human activities determined with terrain conditions were considered as a major contributor to the spatial variations.In addition,the biomass along the north to south gradient showed certain increasing trend which might indicate that environmental factors such as nitrogen deposition have exerted impacts on forest development.The efforts of using tree ring documents to evaluate remote sensing derived results also yield positive outcomes.Landsat derived biomass estimations showed good correlations with the results reconstructed from tree ring width series.However,coefficient of determination(R2)decreased as the date went back to 1990.This phenomenon might mean that the errors in the remote sensing derived results would gradually increase for previous periods.In addition,regression analysis also found that remote sensing based method might have underestimated the overall biomass levels.In summary,LiDAR data could provide high precision forest aboveground biomass estimations,which could effectively support regional biomass estimations using optical remote sensing images.Meanwhile,the tree-ring data provided important information on forest growth,which could be used for validating long-term biomass estimations,although the verification method should be further improved in future research.Additionally,the thesis argues that annual growth information imbedded in tree-rings and the environmental change information that could be determined by tree ring chemistry analysis would help to interpret the spatiotemporal variations of biomass created with remote sensing methods.The underlying mechanisms could be analyzed as well.This could be an important avenue for next step researches.