中文摘要：

针对洞庭湖区,以30m分辨率的环境减灾卫星CCD影像为参考,对比分析了归一化差异植被指数（ND-VI）和比值植被指数（RVI）应用于MODIS遥感影像水体面积提取的优缺点。研究发现,选取NDVI并且赋予各月份各自适应的阈值进行水体面积提取,有效避免了采取单一阈值造成的枯水期水体误提以及丰水期水体漏提问题。将提取湖区面积与相应日期水位数据组成水位-面积组,建立湖区面积-水位相关关系。考虑到建立湖区面积与单一水文站点水位间相关关系存在一定空间不合理性,选取逐步多元回归法建立2003～2006年湖区面积与多站点水位间相关关系。结果表明：在高水位和低水位处,洞庭湖面积-水位关系年际间变化不太明显,但在中等水位处（如24～29m）,湖区面积-水位关系有比较明显的变化,同一水位处湖区面积有逐年减小的趋势。

英文摘要：

Abstract; NDVI and RVI are applied to extract water body information of Dongting Lake from MODIS ima- ges,and the accuracies are assessed with CCD images chosen as reference. The NDVI model is selected and different thresholds are set for the different months, which can effectively avoid extracting other features as water body by mistake in dry season and can also resolve the problem of the missing of water body in wet season. With the statistics of the water area of Dongting Lake and the corresponding observed daily average water levels, it is available to establish the relation of the water area of Dongting Lake to the water levels. In consideration that deriving the lake area from the water level information of a single station is not fully reasonable because of the uneven distribution of water levels among different hydrological stations, the formulas between the lake area and the water levels of multiple stations for the period of 2003-2006 are established by stepwise multiple regression method with SPSS. After validated by F-test and t-test, the regression equations and the regression coefficients are both extremely significant and can be used for forecast. Conclusions from the study are made as follows. There is no significant variation in the lake area- water level relationship at high or low water levels for the period of 2003-2006, but for a certain water level between 24 m and 29 m, the area of Donting Lake has a trend of decrease.