Evaluation of Chlorophyll-a estimation using Sentinel 3 based on various algorithms in southern coastal Vietnam

• Published in: International journal of applied earth observation and geoinformation Vol. 112; p. 102951
• Main Authors: Binh, Nguyen An, Hoa, Pham Viet, Thao, Giang Thi Phuong, Duan, Ho Dinh, Thu, Phan Minh
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.08.2022; Elsevier
• Subjects: Bio-optical algorithms; C2RCC; Chlorophyll-a; DINEOF; DSF; OLCI; Sentinel 3; DINEOF; Chlorophyll-a; OLCI; Bio-optical algorithms; DSF; Sentinel 3; C2RCC
• ISSN: 1569-8432; 1872-826X
• DOI: 10.1016/j.jag.2022.102951

•Both satellites Sentinel 3A and 3B were assessed as well as the synergy between them.•Case 2 Regional CoastColour (C2RCC) and dark spectrum filling (DSF) were used for atmospheric correction schemes.•Validation with in-situ data collected in different locations and periods.•OC4ME and OC5 is the most accurate chlorophyll-a algorithm compared to OC4 and OC6.•Data Interpolating Empirical Orthogonal Functions (DINEOF) can overcome cloud contamination to retrieve gap-free spatial chlorophyll-a concentration products.•Results demonstrated an adequate solution for cloudy coastal Vietnam based on Sentinel 3A and/or 3B. This paper aims to assess the potential of Ocean Land Colour Instrument (OLCI) for the retrieval of chlorophyll-a (chl-a) over southern coastal waters of Vietnam. For that purpose, four chlorophyll-a ocean color (OC) algorithms (OC4ME and three new OC version 7 OC4, OC5, OC6) were applied based on water-leaving reflectance obtained from two atmospheric correction processors (C2RCC and DSF). To overcome high cloud coverage in the area of interest, full spatial data reconstruction was implemented using Data Interpolating Empirical Orthogonal Functions (DINEOF). Numerical error metrics of in situ measurements (n = 49) collected in different ship-based campaigns has been assessed for Sentinel-3A (S-3A) and 3B (S-3B) as well as on the combined products built from these two later satellites. Results showed that products based on C2RRC significantly outperformed DSF. For chl-a algorithms, C2RCC-based OC5 gave the most accurate retrieval while applied to S-3A (R2: 0.58, RMSE: 1.018 mg m−3, MAPE: 49.4 %), S-3B (R2: 0.75, RMSE: 0.776 mg m−3, MAPE: 37.3 %), and synergy datasets (R2: 0.70, RMSE: 0.844 mg m−3, MAPE: 42.5 %). With>50 % of observations missing due to cloud cover, DINEOF provides a promising solution to reconstruct the full spatial information. The successfully demonstrated retrieval of chl-a in our study presents potential for daily monitoring when combining observations from S-3A/B to further improve our understanding of the spatio-temporal dynamics of coastal ecosystems.