| Recently, serious forest decreasing, land degradation, environment pollution, biodiversity losing, especially carbon dioxide concentration increasing and greenhouse effect etc. are most crucial global enveironment change problems. Thus carbon cycle study in global scale is paid more attention currently. To assess precisely the role forest playing in global carbon balance and understand the role forest ecology system playing in carbon cycle, it is increasingly gaining the global attention to study the carbon dynamic more accurately. As a base to analyze the terrestrial carbon cycle and dynamic, forest biomass estimation has been one of important contents of ecology and global change study. 3S (RS, GIS & GPS) technique with its development continuously offers a potential method to study biomass in large scale.To meet the demand, taking Wangqing forest area as the experiment base, the study used 3S integrated technique to do investigation in both theory and method so as to estimate forest biomass precisely in real time effectively, and offer the practical experiences to do further research in China. The key contents of the study is as following:(1) Based on the forest inventory, the compatible forest biomass model design concept has been realized. Taking stem volume as one of independent variables, the study produced a series of compatible biomass models for different forest types using the joint equations, and gained quite high accuracy. It has basically solved the problem of compatible forest biomass models, and thus was a great developing in the relative research field. Employing the models, the forest biomass in the study area for conifer forest, broad leaf forest and mixed forest was estimated as 97.78, and 110.44 t·hm-2 respectively, and among them tree biomass accounted for 95.01%, 93.89% and 94.2% respectively for three forest types. For any forest layer including tree, shrub and grass, the relative biomass density in broad leaf forest was highest, followed by that in mixed and conifer forest. The ratio shrub & grass biomass accounted in broadleaf forest is highest, followed by that in mixed and conifer forest.(2) With GIS (Geographic Information System), applying the B-P Nerve Network the nonlinear forest biomass RS (Remote Sensing) modeling system was designed. Except for the RS factors such as Digital Number of bands and Vegetation Index etc, ths study the study adopted as well as some quantified and qualified factors such as Landform type, Altitude, Slope and Aspect and so on. During modeling stage, by certain methods such as reducing input data and enhancing arithmetic of exercise and so on, the standard B-P Nerve Network was developed and applied to model the forest biomass. The result denoted that enhanced B-P Nerve Network had the faster convergence speed and stronger self-study and self-adaptation function, was able to use transcendent knowledge of the defined RS sample data sets and draw automatically the reasonable model. Therefore, the study decided to employ the enhanced B-P Nerve Network to construct a forest biomass modeling system. It was found in the study that the data set after geometric transformation was more suitable than that before geometrictransformation to model the forest biomass. In addition, the data set drawn from 75 m buffer zones was more suitable than that from 15 and 45 m buffer zones to model the forest biomass. For conifer, broad leaf and mixed forest biomass modeling system, the mean relative error of enhanced simulation results was -1.47%, 2.38% and 3.56% respectively, and the mean relative absolute error was 6.33%, 8.46% and 8.91% respectively. Combined the land cover map accuracy (90.47%) together, the final predicted of the enhanced B-P Nerve Network forest biomass model was 88.04%, which was able to meet the production demand. The ideal predicted estimation was achieved and the forest biomass and carbon storage distribution maps were produced at the same time.(3) From the spatial analysis based on the forest biomass map, the following information was drawn. The broad leaf forest biomass/carbon storage accounted for 61% of total forest biomass in the study area, which took a dominant status, and following by that of mixed and conifer forest as 22% and 17% respectively. From the previous RS estimation, the forest area, the forest storage and carbon pool in China during 1995-1999 was 142.6 Mhm2, 25.77 t-hm"2 and 3.68 Gt C respectively. From the study result, the forest cover of the study area was 0.222 2 Mhm2 (excluding cloud and shadow area and Lanjia area) which accounted for 0.156% of the total forest area of China, however the forest carbon pool was 0.012 Gt C in the study area which accounted for 0.326% of total forest carbon pool of China. Moreover, the mean forest carbon storage was 51.84 t-hm" , which was much higher than the average value of that in China. The order forest biomass/carbon storage distribution from large to small following altitude classes change was middle>lower>higher>low>high. The order forest biomass/carbon storage distribution from large to small following slope classes change was slow slope>slope>even slope>steep slope>dangerous slope. On the 10° slope section, forest was able to absorb the light with the highest efficiency.(4) Taking the biomass map as the basic map, the study made out the forest management planning map. The cover area of the thin cutting area, natural protection area and forestation area was 54 037.35, 64 749.6 and 89 887.68 hm2 respectively. The Ecology, Society and Economy benefit will be gained if the plan could be taken into effect.
|
PDF Full Text Request