Taking some water areas in Zhongshan as the research object, 62 sampling points are selected to analyze the spectral characteristics of malodorous black waters and normal waters. According to this characteristic, this study tries to strengthen the differences between malodorous black waters and normal waters by using a new band combination method, and establishes discrimination models HCI on malodorous black waters, which are compared with the traditional NDBWI and BOI index models. Results show that: (1) The recognition accuracy of ratio algorithm is significantly higher than that of difference algorithm. (2) Selecting different water quality samples to draw the box diagram of each index model as an important basis for defining the threshold, the NDBWI threshold is 0.09, the BOI threshold is 0.06 and the HCI5 threshold is 0.4. (3) Using the synchronous monitoring results of surface water quality to evaluate the identification accuracy of malodorous black waters, the Kappa coefficient is the highest in HCI5 (0.876) and BOI (0.843), followed by NDBWI (0.739) and HCI4 (0.608). (4) The BOI and HCI5 index models are applied to other remote sensing images of Zhongshan City, and they can still distinguish malodorous black waters from normal waters, and have certain universality. Accordingly, this study suggests that BOI and HCI5 should be used as index models for remote sensing identification of malodorous black waters in this area.