湖盆数据未知的湖泊动态库容遥感监测方法

Abstract

针对现有库容遥感监测方法对无湖盆数据区域的湖泊动态库容难以直接测算问题，提出了未知湖泊水下地形数据的遥感湖泊动态库容监测方法。该方法通过多源遥感数据，匹配相对时相的湖泊面积和水位信息，构建并模拟湖盆DEM数据，据此来估算湖泊的动态库容。在算法实现上，首先采用分布迭代水体提取从遥感影像提取湖泊的多期动态边界；其次，从ICEsat GLAS激光测高数据中反演出湖泊的动态水位高程；第三，依据时间水位信息，通过邻近时相匹配，将水位高程赋给湖泊边界线，生成湖泊等水位线；第四，通过等水位线构建TIN(triangulated irregular network)和Kriging 插值，得到模拟湖盆数字高程模型；最后，依据模拟湖盆DEM和水体面积分布、水位信息，计算湖泊动态库容。试验通过对博斯腾湖的多年动态库容监测与真实性检验，结果显示:最大误差为2。21×；108 m3，最小误差为0。000 02×；108 m3，平均误差为0。044×；108 m3，均方根为0。59，相关系数达到0。99。

Alternate abstract:

Focusing on lake underwater terrain unknown and dynamic storage that is difficult to obtain by the traditional methods, a new method is proposed for lake dynamic storage estimation by multi-source and multi-temporal remote sensing without underwater terrain data. The details are as follows. Firstly, extraction dynamic lake boundary through steps by steps adaptive iteration water body detection algorithm from multi-temporal remote sensing imagery. And then, retrieve water level information from ICESat GLAS laser point data. Thirdly, comprehensive utilizing lake area and elevation data, the lake boundary is converted to contour of water by the water level is assigned to the lake boundary line, according to the time and water level information. Fourthly, through the contour line construction TIN (triangulated irregular network) model and Kriging interpolation, it is gotten that the simulated three-dimensional lake digital elevation model. Finally, on the basis of simulated DEM, it is calculated that the dynamic lake volume, lake area distribution and water level information. The Bosten lake is selected as a case studying to verify the algorithm. The area and dynamic water storage variations of Bosten lake are detected since 2000. The results show that, the maximum error is 2.21× 10⁸ m³, the minimum error is 0.00002× 10⁸ m³, the average error is 0.044×10⁸ m³, the root mean square is 0.59 and the correlation coefficient reached 0.99.