| Forest is an important part of terrestrial ecosystem. It exchanges substances and energy with the atmosphere through photosynthesis and evapotranspiration. It can adjust the amount of carbon dioxide in the air, control the climate change and plays an important role to the balance of the carbon cycle. As the research on global change and carbon cycle continued to heat up, it has become the urgent needs for development of global change science and ecological construction to access to a wide range, accurate and detailed information on forest canopy height and biomass. Forest canopy height and biomass is the necessary information of carbon cycle research. The conventional way of accessing forest parameters is through the field investigation, the labor intensity of which is big and this method can only obtain data of some point but will not be able to timely access to a wide range of forest parameters of the spatial distribution information. The application and promotion of remote sensing technology made up for the inadequacy of traditional forest survey and implement the efficient and accurate monitoring of a wide range of space and time to a certain extent.Xiao Xing an Mountains in Heilongjiang province is taken as the study area in this study. The slope correction is conducted in view of flare data in the complex terrain to achieve ICESat/GLAS waveform processing and parameter extraction. The spatial distribution of ICESat/GLAS waveform data are manifested as discrete points, which have no quality of imaging. The research attempts to collaborate with multi-angle optical remote sensing data MISR, Landsat time series data and the big footprint Lidar data for inversion of forest canopy height and biomass to realize the forest graphics with different spatial resolution. The combined use of multi-source remote sensing data and maturely complement of information can provide help for accurate estimation of forest biomass and carbon reserve. The research content is as follows:(1) The canopy height of GLAS waveform data in Xiao xing an ling Mountains is extracted by using method of Gaussian low-pass filtering. Precision evaluation of GLAS forest canopy height is conducted according to cumulative grade group (0-5°,5°-10 °,10 to 15) by using airborne Lidar data. The study found that GLAS data can more accurately deliver the information of forest canopy height and the accuracy of forest canopy height interpretation is 74.7%-74.7%. With the increase of slope, the explanation ability of GLAS waveform data reduced. Slope correction can obviously increase the accuracy of GLAS parameter, but with the increase of slope, correction effect of slope correction declines (5-10 °:2.4% increased by, 10-15°:0.3% increased).(2) The waveform parameters is extracted on the basis of GLAS waveform data processing, and then GLAS flare biomass model is set up by respectively using linear stepwise regression algorithm and Erf_BP neural network algorithm. Compared the performance of the two models, predictive ability of Erf-BP neural network model (P= 0.965, RMSE-0.965 t/ha) is superior to linear stepwise regression model (P= 0.86, RMSE= 0.86 t/ha). Sample size of forest aboveground biomass in the study area can be accurately and effectively enlarged by using biomass inversion of Erf-BP neural network model to GLAS flare in study area, which can provide sufficient data for a wide range of biomass research.(3) Taking GLAS laser points as sample data, tree height and biomass model is set up by using the multi-angle and multispectral reflectivity information of MISR data to realize the seamless inversion of forest tree height and aboveground biomass in large space and large scale.(4) In the local area, Landtrendr calculation method is tried to be used to extract the forest history disturbance and repairing information of landsat time series data and explore the estimation ability of forest history disturbance and repairing information to forest tree height and aboveground biomass. |
PDF Full Text Request