Evaluation of chlorophyll-a retrieval algorithms based on MERIS bands for optically varying eutrophic inland lakes

• Main Authors: LYU, Heng, LI, Xiaojun, WANG, Yannan, JIN, Qi, CAO, Kai, WANG, Qiao, LI, Yunmei
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2015
• Subjects: Accuracy evaluation; Chlorophyll-a concentration; MERIS bands; Optical clustering; Retrieval algorithms; Databases and Information Systems; Theory and Algorithms
• Online Access: https://ink.library.smu.edu.sg/sis_research/5444
• Institution: Singapore Management University

Fourteen field campaigns were conducted in five inland lakes during different seasons between 2006 and 2013, and a total of 398 water samples with varying optical characteristics were collected. The characteristics were analyzed based on remote sensing reflectance, and an automatic cluster two-step method was applied for water classification. The inland waters could be clustered into three types, which we labeled water types I, II and III. From water types I to III, the effect of the phytoplankton on the optical characteristics gradually decreased. Four chlorophyll-a retrieval algorithms for Case II water, a two-band, three-band, four-band and SCI (Synthetic Chlorophyll Index) algorithm were evaluated for three water types based on the MERIS bands. Different MERIS bands were used for the three water types in each of the four algorithms. The four algorithms had different levels of retrieval accuracy for each water type, and no single algorithm could be successfully applied to all water types. For water types I and III, the three-band algorithm performed the best, while the four-band algorithm had the highest retrieval accuracy for water type II. However, the three-band algorithm is preferable to the two-band algorithm for turbid eutrophic inland waters. The SCI algorithm is recommended for highly turbid water with a higher concentration of total suspended solids. Our research indicates that the chlorophyll-a concentration retrieval by remote sensing for optically contrasted inland water requires a specific algorithm that is based on the optical characteristics of inland water bodies to obtain higher estimation accuracy.