Analysis of the use of NDWI_green and NDWI_red for inland water mapping in the Yellow River Basin using Landsat-8 OLI imagery

ABSTRACT

Clear and green water bodies reflect more green light than red light, while brown water bodies reflect more red light than green light. It is thus believed that the normalized difference water indexes using the green band and infrared wavelengths (NDWI_green) and the normalized difference water indexes using the red band and infrared wavelengths (NDWI_red) would act differently in mapping inland water bodies with different colours. This study aims to evaluate the performance of NDWI_green and NDWI_red in mapping water bodies with different colours in the Yellow River Basin using the Landsat-8 Operational Land Imager (OLI) imagery. The results show that NDWI_green and NDWI_red achieve overall accuracy of 95.99–96.82% and 88.32–96.68% in clear water mapping, 95.58–96.35% and 93.50–94.60% in green water mapping, 88.12–92.63% and 90.08–93.10% in brown water mapping, respectively. It is shown that both NDWI_green and NDWI_red can serve the purpose of inland water mapping. Moreover, the comparative analysis suggests that NDWI_green is more suitable for clear and green water mapping, while NDWI_red is more adequate for brown water mapping.

Acknowledgments

The authors are indebted to the USGS for free imagery. We are grateful to Baolin Zhu for his help in collecting water quality data. We also want to thank the reviewers for their constructive comments which help improve this paper significantly.

Additional information

Funding

This study was financially supported by the Opening Project of Key Laboratory for the ecology and environment of river wetland in Shaanxi Province [Grant number SXSD201701], Natural Science Foundation of China [Grant number 41071271 and 41501460] and Shaanxi Province Foundation Research Project [Grant number 2015JM4132].