Comparison of fCO2 trends in river dominant and ocean dominant ocean margins

Surface ocean fCO₂ often exhibits large fluctuations and heterogeneity because of multiple controlling factors, which pose a challenge for long-term trend analysis, especially in the ocean margins. We propose a new statistical approach, Generalized Additive Mixed Models (GAMM), to interpret oceanic fCO₂ trend in two different ocean margins, i.e., an ocean-dominated margin located in the Northwestern Pacific Ocean near Japan (1981-2014, 130-148°E, 28-43°N), and a river-dominated margin in the Northern Gulf of Mexico (nGoM) (2003-2014, 130-148°W, 27.5-30.4°N). Using the GAMM method, we were able to derive long-term fCO₂ trends with both improved precision and greater robustness to data gaps compared to a classic deseasonalization method in the literature. The fCO₂ trend in the Japan margin was 2.0±0.6 µatm yr^-1 (mean±standard deviation, n=99), which was very close to the atmosphere fCO₂ trend (1.90±0.02 µatm yr^-1). To distinguish the driving forces for the fCO₂ trend in the river-dominated margin, we divided the nGoM into two zones: low salinity (0<s≤25) and high salinity (s>25). The fCO₂ trend in high salinity zone was 1.5±0.7 µatm yr^-1 (mean ± standard error), whereas the fCO₂ trend in low salinity zone showed an opposite trend (-23±5 µatm yr^-1). We proposed that the decreasing surface fCO₂ in the river-dominated ocean, in particular the plume area, is a result of enhanced primary production due to eutrophication; while fCO₂ trend in ocean-dominated margin was mainly controlled by atmosphere CO₂ increase.