A weighted normalized difference water index for water extraction using Landsat imagery

ABSTRACT

Water extraction is one of challenging topics in studies on remote-sensing applications. Spectral profiles and experiments indicate that existing water indices often misclassified turbid water, small waterbodies, and some land features in a shadow area. In this study, a new water index called weighted normalized difference water index (WNDWI) was proposed to reduce those errors and improve the mapping accuracy of waterbodies by using Landsat imagery. To test the performance of the newly proposed water index, two test sites (Tampa Bay, FL, USA and Xiangshan Harbour, Zhejiang, China) were selected and the performances of three existing water indices including the normalized difference water index (NDWI), the modified NDWI (MNDWI), and the automated water extraction index (AWEI) were compared with that of the WNDWI. In addition, a default threshold 0 and automatically thresholding methods including Otsu threshold method and multiple thresholds identified by valley points in a histogram curve were tested to determine an optimal threshold that can be used to separate water and non-water features from grey images created by the four water indices. The experimental results indicate that the overall accuracies (OAs) created with WNDWI were all higher than those created with the three existing water indices: NDWI, MNDWI, and AWEI in both sites. Moreover, the results thresholded by 0 owned or shared the highest OAs with the results segmented by some of non-zero thresholds obtained from Otsu method and multiple thresholds method. Therefore, using an appropriate threshold, the proposed method could extract waterbodies from Landsat TM imagery with a high accuracy.

Acknowledgement

This work was supported by National Natural Science Foundation of China: [Grant Numbers U1609203 and 41471004].

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China: [Grant Numbers 41471004 and U1609203].