Seasonal Forecasts of pH and Aragonite Saturation for the Bering Sea Shelf

Abstract

The number of numerical model forecasts of ocean environmental conditions has greatly expanded in recent years, including biogeochemical variables. Forecasts can provide marine resource managers with advance warning of extreme events such as heatwaves and hypoxia, though forecast products are scarcer in high‐latitude environments. The Bering Sea shelf is a large marine ecosystem that supports critical commercial, cultural, and subsistence ecosystem services that are vulnerable to extreme events and anthropogenic stressors such as ocean acidification (OA). Here, we use a regional oceanographic model of the Bering Sea to assess model forecast skill in predicting bottom water pH and aragonite saturation state (Ω _arag ) on lead times from 1 to 9 months. We simulate 28 years (1982–2010) of 3‐member ensemble retrospective forecasts, initialized both in April and May following the retreat of winter sea ice and several months in advance of early fall when the most acidic bottom water conditions occur. The results suggest that the model is skillful (anomaly correlation coefficients > 0.5) in forecasting shelf‐wide anomalies in bottom water pH and Ω _arag , on lead times up to 9 months owing primarily to strong persistence. However, predictive skill for finer scale spatial anomalies is more limited and variable. Model forecast skill is also strong in Bristol Bay, home to the red king crab fishery that has faced recent closures and is threatened by OA. The model forecasts presented here can provide several months advanced notice for acidified water conditions and bolster a suite of products used to support evidence‐based fisheries management.