针对MODIS数据绿潮提取存在大量混合像元导致绿潮覆盖面积偏大这一问题, 基于3 m分辨率的机载SAR高分影像, 结合归一化植被指数(NDVI), 对MODIS研究区中大于NDVI阈值的像元进行混合像元分解得到绿潮的“覆盖面积”; 获取的准同步3 m机载SAR提取的绿潮面积为“真实值”, 然后建立二者之间的关系模型, 并选取不同的样本区域对该模型进行了验证。实验结果表明: NDVI等传统算法所提取的绿潮覆盖面积约为“真实值”的2.68倍; 基于混合像元分解的方法所提取的绿潮面积较“真实值”偏小, 约为“真值”的0.56倍; 与传统的NDVI等多波段比值法相比, 该精细化模型方法提取的绿潮覆盖面积更接近于“真实值”, 与“真实值”误差仅为6.7%。

MODIS’s existence of a large number of mixed pixels caused the larger coverage area of green tide. For this problem, based on NDVI algorithm, we acquired green tide’s covering area by mixing pixels’ decomposition which threshold value is larger than NDVI. Setting the quasi-synchronous green tide’s covering area extracted by 3 m airborne SAR data as “true value”, we established the relationship model between the two to realize the refined extraction. The experimental results suggested that: the covering area extracted by NDVI algorithm is about 2.68 times than “true value”; the covering area extracted by mixing pixels’ decomposition is lesser than “true value”, about 0.56 times compared with “true value”; the covering area extracted by the refined method, compared with traditional multi band ratio algorithm such as NDVI, was closer to the “true value” with an error of 6.7%.