| The Pearl River Estuary wetland in the developed region was selected as the studying object in this thesis. The integrated approach which including the remote sensing and geography infor-mation system(GIS), the field survey, and collecting history data was adopted to analyze the spa-tio-temporal evolution in recent twenty years and it's driver force, some countermeasures for regulating and controlling were put forward. Based upon the hotspot problem, which is the mechanism of the wetland landscape pattern evolution and its drive force analysis in the global change background, the research can not only provide a new angle of view and important gist for Pearl River Estuary wetland protection and degraded wetland ecosystem resume and wetland landscape planning, but also arrive to the goal of revealing the mechanism of estuary wetland landscape evolution and establishing the drive force model and countermeasures for regulating and controlling, also provide some demonstration and reference for other estuary wetland.In the wetland classification of east and west shore of inner Lingding ocean, taking ERDAS software as the flat roof, gradation and classification method based-on knowledge were adopted. The expert system with the combination of geography information and comparing after selecting classification were used in mangrove wetland classification. In the extraction of the dike-pond wetland information, dike-pond information was created after PCA transition of TM5 image and image difference transaction using the spectral character. The landscape pattern indexes were cal-culated using FRAGSTATS grid-edition. The qualitative and quantitative methods were used in the analysis of drive force. Taking Qi'ao island as an example, using artificial neural network to study the mangrove forest community classification. The estimation models of mangrove vegeta-tion biomass were established through combination of photics remote sensing and radar remote sensing after regression analysis of remote sensing images and synchronization actuality biomass investigation.The main conclusions were summed up as following:(1) It is great advantage to study the wetland landscape evolution using remote sensing and GIS in the regional scale of the Pearl River Estuary. The gradation and classification method based-on knowledge can overcome the shortage of human-computer alternation translation with eyes, realize the goal of classification for great region and the more temporal images, and also ful-fill the landscape mapping. In this paper, the total classification resolution reach 85 percent, the coefficient of kappa exceed 0.8.(2) Under natural disturbance and man made disturbance, wetland landscape pattern of Pearl River Estuary in recent 20 years changed evidently. The results indicated that the areas of wetlands including river, reservoir, dike-pond, tidal flat, and mangrove forest reduced 12270 hm2 during the period of 1988 to 2004. The proportion of wetland accounting for the research area descended from 43.4 percent in 1988 to 40 percent in 2004. The land use conversion matrix analysis indi-cated that 91.92 percent tidal flat, 80.43 percent dike-pond, 62.21 percent mangrove forest, 54.63 percent reservoir, and 13.06 percent river transferred into other land utilization from 1988 to 2004. Mangrove area of the Pearl River Estuary decreased early but increased later, from 457.2 hm2 in 1988 to 416.9 hm2 in 2002, which were mainly distributed in Dawei bay, Shenzhen bay and inner Lingding ocean. Dike-pond area of research region increased 13.2 percent in recent twenty years.The estuary wetland landscape has been changed very distinctness. The contrived disturbance become the main drive force of the landscape evolution. In recent 20 years, total patches, diversity index and evenness index of the research region increased early but decreased later, average patch size and aggregation index decreased early but increased later, division index and separate index increase. The estuary wetland landscape has been influenced by ecological activities such as city construction, sea shore water conservancy, inning, aquiculture, etc.(3) To attempt to study the mangrove forest community classification using artificial neural network and to study the biomass using remote sensing. Using the established model to estimate the mangrove vegetation biomass may not only provide important gist for wetland protection and forest pattern optimization, but also provide important information for simulating the global change model.(4) The regulating and controlling of the wetland landscape in the Pearl River Estuary need theories supervision such as landscape ecology, resume ecology, natural cybernetics. The govern-ment interference on tidal flat wetland inning under uncertainty, application of wetland vegetation growing model in restoration of mangrove wetland, the influence of irreversibility and uncertainty on the timing of wetland environmental policy, to carry out environment impact assessment for wetland resource and environment, to introduce wetland culture into wetland protection, to use RS, GIS and GPS techniques to monitor wetland evolution, and to pay more attention to the estuary evolution were put forward in the thesis.There are three innovation spots as following:(1) Adopting the different method of classification in different type of wetland may enhance the classification precision. In the mangrove forest community classification, the satisfied results were obtained for the amalgamation image of TM and SAR using ANN.(2) The landscape metrics in the class-level and landscape-level of the studied area in four temporal are calculated. A remote sensing estimation model of mangrove biomass was established.(3) The countermeasures for regulating and controlling were undergone using the qualitative and quantitative methods.
|
PDF Full Text Request