Skillful multiyear predictions of ocean acidification in the California Current System

Abstract

The California Current System (CCS) sustains economically valuable fisheries and is particularly vulnerable to ocean acidification, due to the natural upwelling of corrosive waters that affect ecosystem function. Marine resource managers in the CCS could benefit from advanced knowledge of ocean acidity on multiyear timescales. We use a novel suite of retrospective forecasts with an initialized Earth system model (ESM) to predict the evolution of surface pH anomalies in the CCS. Here we show that the forecast system skillfully predicts observed surface pH variations multiple years in advance over a naïve forecasting method. Skillful predictions of surface pH are mainly derived from the initialization of dissolved inorganic carbon anomalies that are subsequently transported into the CCS. Our results demonstrate the potential for ESMs to provide predictions relevant to managing the onset and impacts of ocean acidification in this vulnerable region.

Predictable variations of the carbon sinks and atmospheric CO2 growth in a multi-model framework

“…Longer predictability of 4-7 years in regions like the North Atlantic and the Southern Ocean is suggested (Li et al, 2016;Fransner et al, 2020). ESM-based initialized prediction systems also demonstrate predictability of other marine biogeochemical properties such as net primary production, export production, and seawater pH (Brady et al, 2020;Fransner et al, 2020;Krumhardt et al, 2020;Park et al, 2019;Séférian et al, 2014;Yeager et al, 2018). On the land side, a potential prediction skill of 2 years was established for terrestrial net ecosystem production (Lovenduski, Bonan, et al, 2019), but only of 9 months for tropical land-atmosphere carbon flux (Zeng et al, 2008).…”

Physical and Biogeochemical Drivers of Alongshore pH and Oxygen Variability in the California Current System

“…Coupled physical‐biogeochemical models yield insight into the mechanisms shaping alongshore variability at spatiotemporal scales that observational studies cannot concurrently resolve. Existing modeling studies for the CCS have provided important knowledge about processes controlling pH and oxygen variability, and the predictability of their seasonal, interannual, and long‐term evolution (Brady et al, 2020; Dussin et al, 2019; Gruber et al, 2012; Hauri, Gruber, McDonnell, & Vogt, 2013; Hauri, Gruber, Vogt, et al, 2013; Howard et al, 2020; Siedlecki et al, 2015, 2016). Many of these studies focused on regional‐ and basin‐scale mechanisms influencing ocean acidification and hypoxia, obscuring pH and oxygen variability at finer alongshore scales (10–100 km) over which processes modulating these variables (e.g., upwelling intensity, nutrient transport, and primary production) are known to vary in the central CCS (Fiechter et al, 2018).…”

Assessing Decadal Predictability in an Earth‐System Model Using Explainable Neural Networks

“…The discovery of this coupling has led to the usage of oceanic variability to make decadal predictions of atmospheric anomalies relevant to society. Recently, oceanic observations have been assimilated into Earth‐system models to generate large ensembles of global decadal predictions (Meehl et al., 2009; van Oldenborgh et al., 2012; Yeager et al., 2018), which have a reasonable amount of prediction skill for variables such as continental temperature and precipitation (Smith et al., 2019) and ocean acidification (Brady et al., 2020). Additional efforts have created statistical decadal prediction models based on knowledge of specific modes of oceanic decadal variability (e.g., Simpson et al., 2019).…”

“…of prediction skill for variables such as continental temperature and precipitation (Smith et al, 2019) and ocean acidification (Brady et al, 2020). Additional efforts have created statistical decadal prediction models based on knowledge of specific modes of oceanic decadal variability (e.g., Simpson et al, 2019).…”