| The forest area is decreasing year by year, but bamboo as a special forest type,its area is growing at a faste rate, especially for moso bamboo’s high carbon sequestration potential, their great carbon storage contributing to global carbon balance has been gotten wide attention and gradually been recognized. Some researches were focused on bamboo forests monitoring, biomass/carbon estimation, and bamboo forest structure parameters inversion.However, these works only used Landsat TM data and were at the county level. China lies in the center of bamboo distribution and has huge bamboo forest area, single Landsat TM data source is difficult to achieve large scale synchronous monitoring of bamboo forests and carbon storage.Study area, located in the Yangtze River Delta region of China includes northeastern part of Zhejiang province, southeastern part of Jiangsu province and Shanghai city. The objective of this research is to extract bamboo forests information using MF technique and then estimate carbon storage of bamboo forests based on the combined use of Landsat TM and MODIS data sources. To do this, this research will provide a method for large scale bamboo forests detecting and more completely evaluate contribution of bamboo forests to carbon sequestration. Mainly included the following aspects:1.Carbon storage of each plot was calculated based on ground investigation, bamboo biomass model and conversion factor between biomass and carbon storage was used when Carbon storage of eavh plot was converted to MODIS scale according to area ratio.2. Data of MODIS and TM thematic information were registrated and cut, TM thematic data was converted to MODIS scale.3.The first five MNF components explained over90%of the variance in the original MODIS data and were selected as feature variables and classified into five types using the maximum-likelihood method, included forest, water, farm land, city and mudflat.4.Matched filtering method was used to select the moso endmember and get relative abundance, TM thematic data after scale convert was used to correct the relative abundance and to get pixel abundance of moso banboo.5.Backward stepwise procedure was used to select independent variables, linear regression model was built to estimate moso bamboo forest AGC. Plot data was used to estimate the model, spatial distribution of carbon storage was estimated based on that. This study mainly gets the following conclusion:1.Moso bamboo area of Anji was used to evaluate the TM thematic data’s scale convertation accuracy, the correlation coefficient was0.9975and standard deviation was821.1according to analyze the distribution of Moso bamboo.2.Study area was classified into water, mudflat, farmland, city and forest.Total classification accuracy of Modis data was92.97%, and the kappa coefficient was over0.888, The classification result is suitable for application.3.Band of MNF2and MNF3was used to extracted moso endember, result of matching filter shows that most countries have high precision. According to the corrected moso bamboo abundance, relation of estimated area and actual area R2was0.8453. moso bamboo forest information extraction precision was78.99%in contury scale and96.66%on province scale, but there were still some area was overestimated such as Chunan,Tonglu,Xianju.4.Variables of NDVI2, MNF3, MNF5, MNF9, MNF10were selected to build moso bamboo estimation model, adjustment coefficient R2was0.434, correlation coefficient was0.701. Test of Linan plot data showed that R2between carbon reserves prediction was0.4618, prediction accuracy was87.73%,shows the model has good prediction ability. Storage density distribution map of moso bamboo carbon was mde, we can see that the carbon storage of most conturies lies between0and15Mg C/ha. The results can reflect the practical situations of the study area. |
PDF Full Text Request