摘要
利用统计分析的方法,根据2001年获取的13景鄱阳湖区无云MODIS影像中的9景提取的水体面积,并结合同步观测的水文数据分别采用线性、对数和指数3种模型模拟湖面积-水位之间关系。结果显示对数模型相关性最好(R2=0.918),其次为线性和指数模型。利用另外4景MODIS影像对模型进行检验表明,该模型精度较高,模拟的最大误差为3.36%。本研究显示,可根据鄱阳湖水位观测值,利用该模型预测鄱阳湖洪涝期洪水淹没面积,以弥补云天状况下光学遥感难以监测到洪水淹没范围的不足。本研究为利用遥感影像实时监控鄱阳湖水情空间动态变化提供了可行的方法,对湖泊、水库的泛洪监测、调洪功能分析具有重要意义。
As the largest fresh lake in China, Poyang Lake is also the place where flooding is frequently occurred. Flooding is the major disaster that hinders the development of this area. It's essential to monitor the flooding water progress before the disaster could be prevented and a quick, accurate, and automatic extraction method of water body is of great importance for this purpose. Furthermore, understanding the relationship between water level and flooding water area of the Poyang Lake is critical for flooding prevention, disaster reduction, flooding risk assessment, and disaster assessment in this area. In this paper, we attempted to study the relationship between water level and water coverage area of the Poyang Lake area by using MODIS images and gauging data, and visualize the process of flooding progress with support of remote sensed data. 13 scenes of cloud-free MOIDS images and lake water level data of 2001 were obtained and were used to examine the relationship between water level and lake area. NDVI ( Normalized Difference Vegetation Index) images of the study area were derived from the acquired MODIS images. Water body was then extracted from NDVI images by setting a proper threshold to separate water body from other features. Lake area of different months in 2001 was then calculated based on the results of water extraction from 9 scenes of images. The corresponding water level data of the same time were collected from gauging station. Exponent, linear, and logarithm models were employed to correlate water level against lake area. The results showed the correlation coefficient R^2 are 0. 849,0. 876, and 0. 918, respectively. It was suggested that the lake area is logarithmically related to water level. Another 4 scenes of MODIS images were used to validate the model. The results showed the logarithmical model works well as to predict the water coverage area with water level data and the errors are all less or equal to 3.36%. The major error source lies in the insufficient samples used to derive the model. The findings could be used to predict flooding area when lake water level can not be observed by remote sensed image. From this study,we can come to the conclusion that MOIDS images are valuable data source for accuratly extracting water body and providing vivid visualization of flooding progress. We also conclude that there exists logarithmical relationship between lake area and lake water level in the Poyang Lake area.
出处
《第四纪研究》
CAS
CSCD
北大核心
2008年第2期332-337,共6页
Quaternary Sciences
基金
国家重点基础研究发展规划项目(批准号:2003CB415201)
国家自然科学基金项目(批准号:40672206)资助