The augmented normalized difference water index (ANDWI) is novel and robust. However, in some cases, ANDWI may be less effective in extracting water body with more surface-floating algal bloom. Therefore, considering the characteristics of ANDWI and the construction of the improved eutrophic water body index (MEWI), this paper constructed the modified augmented normalized difference water index (MANDWI) based on the spectral characteristics of eutrophic water body and other surface features. In this paper, water extraction verification was carried out in Lake Taihu with different seasons and different algal bloom levels. MANDWI was compared with DIBWI (double infrared band water index), MEWI, MBCWI (multi-band combination water index), and ANDWI. The results showed that MANDWI was suitable for different seasons and had a good water extraction effect. With the increase of algal bloom level, MADNWI still has high water extraction accuracy (overall accuracy is 0.99, Kappa coefficient is 0.98), while the extraction effects of other indices were poor. The proposal of MANDWI enhanced the ability of remote sensing water extraction, which was of great significance for water resource management and protection, and also provided a reference for the development of similar water body index.