Assessment of chlorophyll-a using multiple and ratio bands algorithm in Sentinel-2: A case study in the coastal areas of Semarang

Remote sensing techniques have been widely used to monitor chlorophyll-a (Chl-a) concentrations in coastal waters. However, the accuracy and transferability of retrieval algorithms across different coastal environments remain uncertain. In this study, we evaluated the performance of a Chl-a retrieval algorithm originally developed for other coastal systems and recalibrated it using high spatial resolution (10 m) Sentinel-2 A (S2A) imagery. The MR4B algorithm employs blue, green, red, and near-infrared (NIR) bands for Chl-a estimation. Validation was conducted against in situ measurements collected on August 21, 2021, July 27, 2022, March 19, 2023, and October 10, 2023, coinciding with S2A overpass times. In addition, a new empirical model, based on green–blue and green–red band ratios, was developed. The MR4B algorithm performed well during the August and July campaigns but exhibited substantial errors for the March and October datasets, with RMSE (MAPE) values of 2.47 μg/L (23.52%), 2.72 μg/L (29.65%), 6.15 μg/L (140%), and 7.23 μg/L (256%), respectively. In contrast, the green–blue band-ratio algorithm produced more stable estimates of mean Chl-a concentrations, with RMSE (MAPE) values of 2.55 μg/L (26.69%), 4.04 μg/L (51%), 4.37 μg/L (64%), and 5.95 μg/L (200%) for the corresponding dates. Overall, the results demonstrate that the S2A green–blue ratio algorithm provides a more robust and operationally practical approach for Chl-a monitoring in this coastal system, owing to its simpler formulation and more consistent performance relative to the MR4B algorithm.