Development of Parameter-Tuned Algorithms for Chlorophyll-a Concentration Estimates in the Southern Ocean
2025
Mingxing Cha | Xiaoping Pang | David Antoine
Accurate estimates of Chlorophyll-a (Chl) concentration from satellite observations are critical for understanding large-scale phytoplankton variations, particularly in the context of climate change. However, existing operational Chl retrieval algorithms have been shown to perform poorly in the Southern Ocean (SO). To address this issue, this study proposed improved Chl algorithms tailored to the SO. To this end, three Chl satellite products (MODIS, OC-CCI, and GlobColour) were evaluated against high-precision (high-performance liquid chromatography-derived, HPLC), long-term (1997&ndash:2021), and spatially widespread (south of 40°:S) in situ Chl observations. Subsequently, OC3M-based empirical algorithms were improved using remote sensing reflectance (Rrs) data. Among the original products, OC-CCI exhibited the best overall performance (R2 = 0.36, Slope = 0.36), followed by GlobColour-AVW (R2 = 0.27, Slope = 0.21), whereas Aqua-MODIS showed the worst agreement (R2 = 0.18, Slope = 0.18) with in situ observations. All three products systematically underestimated Chl concentrations, with average biases of 43% (Aqua-MODIS), 24% (OC-CCI), and 36% (GlobColour-AVW), particularly at high Chl concentrations (>:0.2 mg/m3 for Aqua-MODIS and GlobColour-AVW: >:0.3 mg/m3 for OC-CCI). The parameter-tuned algorithms significantly reduced these biases to 1% (OC-CCI), 3% (GlobColour-AVW), and a slight overestimation of 2% (Aqua-MODIS). All products showed marked improvements in performance, with R2 increasing to 0.68&ndash:0.91, slopes approaching 1.0 (0.62&ndash:0.92), and notable reductions in MAE (1.39&ndash:1.42) and RMSE (1.49&ndash:1.51). These results offer enhanced capabilities for Chl retrieval in the data-sparse and optically complex waters of the SO.
Afficher plus [+] Moins [-]Mots clés AGROVOC
Informations bibliographiques
Cette notice bibliographique a été fournie par Multidisciplinary Digital Publishing Institute
Découvrez la collection de ce fournisseur de données dans AGRIS