鄱阳湖湿地水位与洲滩淹露模型构建

Modeling of water level and submersion or emersion in Poyang Lake wetland

水是控制湿地生态过程的一个重要因子,为了研究洲滩变化和湿地草洲生长发育规律,以便更好地监测和保护湿地生态系统,有必要研究鄱阳湖水体变化特征,分析湿地洲滩水位的时空动态变化和洲滩淹露规律。本文以鄱阳湖国家级自然保护区为例,在对鄱阳湖多年水位进行分析的基础上,利用多时相遥感影像和DEM提取湿地洲滩特征;并在GIS支持下综合考虑地形、鄱阳湖水位、湖泊缓冲区等因素,建立了湿地水位及洲滩淹露模型。时空验证结果表明,该模型具有较好的模拟效果,精度在85%以上。同时,本文根据研究区特点及水体在TM2、7波段的特征差异,构建了一个新的水体指数FDWI,提取水体精度达到98%,特别是对潮湿的沙地、植被和云有很好的区分能力。

Poyang Lake wetland, as an important wetland in the world, is known for its periodic variation of water level in a year affected by the Yangtze River and its tributaries of Ganjiang, Fujiang, Xinjiang, Raojiang and Xiushui rivers. It is very significant to understand the variation of water level for rational exploitation and conservancy of the wetland. This paper, taking Poyang Lake Nature Reserve as a study area, developed a model to simulate the water level and the status of submersion or emersion of wetland using DEM, multi-temporal images and multi-year water level data. First, a new water index FDWI was created to identify the water pixels based on spectrum feature difference between bands 2 and 7 of Landsat TM. The accuracy of classification for water and non-water reaches about 98%. Spatial overlay analysis was performed to obtain the elevations of both water and non-water pixels. Two normal distribution curves were then presented according to elevation statistics. The elevation in the crossed point of two curves was set as the value of water level. With the water level correlation analysis between wetland and Poyang Lake, the regression equations in different lake buffer areas were developed. Furthermore, the model for water level and status of submersion or emersion of wetland was constructed, concerning the topography, buffer areas and water level of Poyang Lake. Finally, the model was validated in estimating the temporal changes of water level of wetland for flood, normal and dry years respectively, as well as in simulating the spatial distribution of both water and non-water pixels of wetland on October 9, 2000. The result shows that the model has satisfactory simulated effect with accuracy above 85 %.