Impact of Temperature on Ocean Color Retrievals through Near-Infrared Iterative Atmospheric Correction Algorithm over Coastal Waters

Accurate atmospheric correction (AC) is vital for ocean color remote sensing, especially in turbid coastal waters where the traditional black pixel assumption in the near-infrared (NIR) bands fails due to substantial water-leaving radiance. This study evaluates the influence of temperature-dependent variability in the absorption coefficient of pure seawater ( a w ) on the widely used NIR iterative atmospheric correction algorithm (AC iter ). A modified algorithm, AC iter-T , is proposed by incorporating temperature-adjusted a w based on empirical measurements. Simulated datasets, covering a wide range of water temperatures, suspended particulate matter ( SPM ), and chlorophyll-a concentrations, were used alongside over 500 satellite–in situ matchups from AERONET-OC sites. Results demonstrate that in turbid waters, especially when the sea surface temperature deviates from the reference (22 °C) by more than 10 °C, the use of temperature-sensitive a w markedly improves retrieval accuracy. The modified AC iter-T algorithm notably reduced remote sensing reflectance ( R rs ) bias and mean absolute percentage difference ( MAPD ) in the blue spectral bands (e.g., 410 nm), with MAPD reductions exceeding 50% in highly turbid and cold water conditions. In contrast, for less turbid waters or when the deviation of water temperature is small, temperature corrections exert minimal influence. These findings highlight the necessity of incorporating temperature-dependent optical parameters into AC frameworks to ensure robust ocean color product accuracy under variable environmental conditions.