| Summary: | 碩士 === 國立屏東科技大學 === 森林系所 === 98 === Kyoto Protocol had declared that forest could mitigate the greenhouse effect of climate change and provided the definition of forest carbon storage and quantitative methods. At present, how to carry out the survey methodology of forest carbon storage is an important issue. Using the sampling ground survey to estimate biomass for regional level is time consuming and difficult to improve the accuracy, therefore, every country were committed to the application of satellite or airborne remote sensing data for estimating the carbon storage to solve the problems of timeliness and accuracy. In forestry, remote sensing provided many kinds of matured technology to use, such as Airborne LiDAR which could detect the variables of stand structure and estimating carbon storage directly or indirectly that was a new application for forest survey in rent years. The 70 years old, different planting density of Cryptomeria japonica experiment area was selected for this study in Chitou area. The timber volume was measured by ground survey and the carbon storage was transferred from the volume and biomass by the conversion coefficient of IPCC method. A small-footprint LiDAR dataset was acquired by Industrial Technology Research Institute (ITRI) over the study area in June 2006. The raw data was processed by the filtering and normalized, and constructed the parameters of stand height, crown base height, echo ratio with CHM. Multiple regression analysis was used for establishment the relationship between the biomass and parameters of estimating from the LiDAR CHM. The results indicated that the LiDAR crown base and 60% stand height were high correlation with biomass (R2=0.70). LiDAR height bins were generated as multiband images of 1m height intervals and 1 m × 1 m pixel dimensions, i.e., 1 × 1 × 1 m voxels. A pixel value represents the number of laser points as a percentage of the total number of points summed up for all pixels in the stack at this position. Average pixel value of different dimension voxels (1 × 1 × 1 m, 2 × 2 × 2 m, 3 × 3 × 3 m, 4 × 4 × 4 m, 5 × 5 × 5 m) were calculated with 15 plots as independent variable for regression with stand biomass. The results indicated that average pixel value of 1×1×1 m voxels of 21, 22, 23, 28 bins were the best fitting with biomass (R2=0.74). In this study, we had two different methods to predicate carbon storage by Airborne LiDAR. How to select adaptive method to estimate the biomass and carbon storage will follow the site situation and status of point cloud distribution.
|
|---|
