Climate models highlight the possibility that the North Atlantic Subpolar Gyre (SPG) could collapse in the next few decades. However, climate models have critical limitations in their ability to correctly represent SPG dynamics and tipping points (TP). Improving models – as required to create an effective early warning system (EWS) – is typically done by deploying new ocean observations and model development projects, but this is time consuming and requires large investments.

SORTED proposes an alternative and more effective approach: to combine recent advances in artificial intelligence (AI) and process-based understanding of SPG TP precursors to extract the required information from existing observations and use these to design an observations-focused EWS. SORTED will demonstrate that delivery of an EWS for the SPG is possible and timely when (a) existing observations are turbo-boosted with AI techniques; (b) innovative EWS techniques based on SPG process understanding from models that feature SPG tipping is applied; (c) the contributions of key processes (e.g. Greenland meltwater, eddies) using high-resolution models are quantified; and (d) proximity to SPG TP is evaluated using statistical and physics-based approaches.

SORTED also emphasizes that new investment in ocean observations for a SPG TP must be informed by a careful evaluation of existing observations. Therefore, in SORTED we will also quantify critical gaps and uncertainties in our EWS design and make recommendations on what and where further observations are needed to provide targets for other groups in time for the second solicitation of this ARIA programme. In doing so, we will address the following programme objectives: 1. Build an EWS through united innovation in observation and modelling. 2. Reduce uncertainty in predictions of when tipping will occur in the SPG system. This proposal addresses Technical Area 3 and is linked to the SORTED concept paper, which was encouraged to proceed to full proposal stage.

Project PI: Dr Alejandra Sanchez-Franks
 

Email: tiadot@noc.ac.uk

Dates
–
Funding

ARIA, Industry.

Aim

SORTED’s novel and ambitious approach to address ARIA’s Forecasting TP challenge will be to push the spatiotemporal capabilities of existing observational records using AI, which when informed by TP knowledge from the models, will make a step change in our ability to detect and monitor early warnings for the SPG TP.

SORTED will thus deliver an EWS for the SPG and reduce uncertainties in TP prediction by first transforming the capacity of existing ocean observations from (a) knowledge of modelbased SPG TP (WP1) and (b) the increase of its spatiotemporal limits using state-of-the-art AI techniques (WP2).

This AI-enhanced observational product will then feed into an upper ocean budget, providing precise quantifications of relative contributions of Subpolar North Atlantic (SPNA) processes to the SPG collapse, together with an assessment of known missing processes in models, such as Greenland Ice Sheet (GrIS) meltwater and the mesoscale field, which will be tested in high-resolution models (WP3), and will be used to determine proximity to a TP using statistical approaches (e.g. [26]) and in-depth SPG process understanding (WP4).

SORTED will also deliver a comprehensive evaluation of uncertainties in our EWS and use it to inform the community of exactly what types of observations are needed and where (WP4), as the evaluation of existing observations is a vital component to future investment in new observations targeting SPG TP.