High-Resolution Neural Network Demonstrates Strong CO2 Source-Sink Juxtaposition in the Coastal Zone

Abstract

The role of coastal oceans in regulating atmospheric carbon dioxide remains poorly quantified and understood. Here, we use a two-step neural network approach to generate estimates from sparse observational data in the coastal Northeast Pacific Ocean at an unprecedented spatial resolution of 1/12° with coverage in the nearshore (0–25 km offshore). We compiled partial pressure of carbon dioxide (pCO₂) observations as well as a range of predictor variables including satellite-based and physical oceanographic reanalysis products. With the predictor variables representing processes affecting pCO₂, we created non-linear relationships to interpolate observations from 1998 to 2019. Compared to in situ shipboard and mooring observations, our coastal pCO₂ product captures broad spatial patterns and seasonal cycle variability well. A sensitivity analysis identifies that the parameters responsible for the neural network's ability to capture regional pCO₂ variability are associated with mechanistic processes, including mixed layer deepening, mesoscale eddies, and gyre upwelling. Using wind speed and atmospheric CO₂, we calculated air-sea CO₂ fluxes. We report an anticorrelation between annual air-sea CO₂ flux and its seasonal amplitude with the relationship driven by circulation, opposing seasonal upwelling/relaxation versus downwelling, and the effects of winter mixing and primary productivity. We show that the inclusion of nearshore net outgassing fluxes lowers the overall regional net flux. Overall, our results suggest that the region is a net sink (−0.7 mol m⁻² yr⁻¹) for atmospheric CO₂ with trends indicating increasing oceanic uptake due to strong connectivity to subsurface waters.