Improved water quality retrieval by identifying optically unique water classes

Majid Nazeer, Janet Elizabeth Nichol

Research output: Journal article publicationJournal articleAcademic researchpeer-review

17 Citations (Scopus)

Abstract

For example, in the coastal waters of Hong Kong water quality varies from east to west. The currently existing water zones, defined by the Hong Kong Environmental Protection Department (EPD) are based on ease of access to sampling locations rather than on water quality alone. In this study an archive of fifty-seven Landsat Thematic Mapper (TM), Enhanced Thematic Mapper Plus (ETM+) and HJ-1 A/B Charged Couple Device (CCD) images over a 13-year period (January 2000–December 2012) was used to define optically distinct water classes by Fuzzy c-Means (FCM) clustering. The clustering was applied by combining the Surface Reflectance (SR) derived from the first four bands of Landsat and HJ-1 scenes with 240 insitu samples of Chlorophyll-a (Chl-a) and Suspended Solid (SS) concentrations collected within 2 h of image acquisition. The FCM clustering suggested the existence of five optically different water classes in the region. The significance of the defined water classes was tested in terms of the water SR behaviour in each band. The SR for Classes 1 and 2 in bands 1–3 was lower than in other classes, and band 4 showed the lowest reflectance, indicating that these classes represent a clearer type of water. Class 3 showed intermediate reflectance in all bands, while Classes 4 and 5 showed overall higher reflectance indicating high sediment contribution from the Pearl River Delta. Application of water quality retrievals within individual classes showed much greater confidence with Root Mean Square Error (RMSE) of 1.32 μg/l (1.21 mg/l) for Chl-a (SS) concentrations, compared with 5.97 μg/l (2.98 mg/l) when applied to the whole spectrum of different water types across the region.
Original languageEnglish
Pages (from-to)1119-1132
Number of pages14
JournalJournal of Hydrology
Volume541
DOIs
Publication statusPublished - 1 Oct 2016

Keywords

  • Coastal waters
  • Fuzzy clustering
  • HJ-1 A/B CCD
  • Landsat

ASJC Scopus subject areas

  • Water Science and Technology

Cite this