27 August 2026
From signals to decisions: Forecasting Tipping Points enters Phase 2
The Forecasting Tipping Points programme is entering its second phase, moving from understanding the science of climate tipping points to building a usable early warning system for the North Atlantic. We will be inviting proposals related to ocean observing, forecasting, and climate-evidence translation as part of a new funding call in September 2026; we’re also keen to hear from prospective users, partners, and co-funders who could help build a lasting capability.
Co-designing an early warning system for the North Atlantic so it is useful, useable and used
Can we recognise when a critical part of the Earth system is approaching irreversible change – early enough for that knowledge to matter? This is the challenge at the heart of ARIA’s Forecasting Tipping Points programme. We launched the programme to test whether observations, physical models and AI could be united into an affordable, sustainable and scientifically credible early warning system for climate tipping points.
The urgency of this challenge is becoming increasingly visible in the North Atlantic. The subpolar gyre – the focus of our early demonstrator – is closely connected to the Atlantic Meridional Overturning Circulation (AMOC), which transports heat through the Atlantic and plays a major role in the climate of Europe and beyond. Here, tipping means a self-reinforcing weakening or loss of deep-ocean convection in the subpolar gyre – an abrupt reorganisation of circulation whose likelihood and scale of impact cannot be assessed without sustained observations of both the ocean and the Greenland ice sheet that helps drive it.
Concern about AMOC is also moving beyond the scientific community. Iceland has elevated potential AMOC collapse as a national security concern; the Nordic Council of Ministers has examined its implications for mitigation, adaptation and governance; Finland has launched government-funded research to understand how weakening could affect its climate and society; and the UK has placed the wider consequences of ecosystem collapse within national security and resilience planning. Financial institutions are also beginning to consider how tipping risks should inform decisions made under deep uncertainty and across different time horizons. This changing landscape reinforces the need to move from describing the risk to building the capability needed to monitor, forecast and respond to it. These changes could affect weather, fisheries, infrastructure and economic security across the North Atlantic, including Greenland, where shifts in ice, freshwater and coastal conditions could directly shape communities, livelihoods and decisions about the future.
Eighteen months in, our Creators are beginning to show what the building blocks of such a system could look like. Teams are assembling new records of the North Atlantic, developing autonomous instruments capable of operating in extreme environments, improving models of the Greenland ice sheet and ocean, and creating new methods to distinguish potential warning signals from the noise and variability of the climate system.
Our teams are also beginning to work as one coordinated programme. Creator-led groups are: connecting hardware, data, models and early warning methods; identifying critical observation gaps; and defining how evidence and uncertainty must move through an end-to-end system.
But this first phase of the programme has also sharpened our view of the problem: an early warning system cannot end with a warning.
Sounding a warning that can change a decision
A warning will have little value if the people who need to use it do not trust the evidence behind it, understand its consequences, or know what action it should trigger.
But who is this early warning system for? Working with the Ditchley and Telescope Foundations, we brought stakeholders from government, finance, fisheries, climate security, and other fields into the programme. These conversations shifted our emphasis from asking when a system might tip and what the consequences would be, to asking what decisions depend on better tipping risk information – and when would the window for making those decisions begin to close.
As a result, we have evolved our thinking about what a successful demonstration must achieve. Improving observations and models remains essential, but Phase 2 of the programme must connect those advances to the people who could use them. Rather than producing a single alarm, we need to show a trusted pathway from a changing physical system, through observations and forecasts, to consequences and proportionate response options.
That means connecting every scientific output to a real pathway from observation to action. What must be measured? How does it improve our estimate of the ocean’s current state? What can be forecast over a decision-relevant timescale? What would that forecast mean for a particular place or sector? Who could act on it, and what level of confidence would they need?
Building a modular early warning system
Phase 2 of the programme will follow a modular approach, organising the programme around an end-to-end early warning system structure.
Modularity matters because no single team or technology can deliver the whole system. It gives each Creator a clearly defined contribution while making the interfaces between observations, models, forecasts and decisions explicit. A modular approach will allow us to test individual components, identify where uncertainty is highest or information is being lost, and improve or replace one part without having to redesign the entire system. It also gives prospective users something tangible to engage with as the early warning capability develops.
We are focusing our early demonstrator on the North Atlantic subpolar gyre, with the Greenland ice sheet as an important source of freshwater entering the ocean. In Phase 1 we focused on building process understanding through separate technical areas and much of that vitally important work will continue. In Phase 2, however, we will bring that work together into an early warning system through six connected modules:
- A sustained and adaptive ocean observing network;
- Measurement of freshwater flowing from the Greenland ice sheet into the North Atlantic;
- Continuously updated estimates of the state of the ocean;
- Detection of physically meaningful early warning signals;
- Interannual to decadal forecasts of plausible ocean change;
- And translation of those forecasts into potential impacts and proportionate response options.
Using a modular approach is intended to make the system buildable and testable with each module having associated cross-programme milestones to achieve. Each component should create something valuable in its own right, while connecting observations, models, forecasts and decisions into an end-to-end early warning architecture will transform existing capability. System integration across the modules will also be a dedicated workstream, where ownership of connecting the pieces will lie with a dedicated team.
Our aim is to move from a compelling concept to a minimum viable product: an early demonstrator that makes the complete data-to-decision pathway visible and allows Creators and prospective users to test what works, what is missing, and what would be required for long-term operation.
Seeking missing capabilities to turn climate observations and models into a usable early warning system
We plan to open a second funding call in mid-September to bring in additional capabilities that are currently missing from our portfolio, or need to be developed at a much greater scale. We expect to seek ambitious proposals in three particularly important areas:
- Ocean observing networks. We are interested in networks, not isolated instruments or short field campaigns. We want creative proposals that address the combination of established international observing systems with lower-cost, autonomous and adaptive technologies to provide sustained, calibrated and year-round coverage of the North Atlantic. We want teams that intelligently direct observations in response to the needs of improved state estimation and forecasting. Co-funding of networks with other entities is encouraged.
- Interannual to decadal forecasting. Decision-makers need insight on timescales over which choices can still be made. In addition to our existing work on long-term projections, we want to advance forecasts on nearer-term timescales of North Atlantic change, combining initial conditions, physical understanding, AI and well-characterised uncertainty. Success will require forecasts that can be tested, compared and improved – not simply long-term projections that cannot be easily verified.
- Impacts and response. Scientific information becomes useful, usable and used when it is framed around a decision. We want teams to translate forecasts into consequences, closing decision windows and proportionate response options for specific users. This work should be co-designed around real decisions, required lead times, acceptable uncertainty and the practical capacity to act.
The strongest ideas may span more than one module. They may also require new collaborations between people who do not normally work together.
Who should be thinking about building with us?
Climate and ocean scientists will be central to this work, but they cannot build an operational early warning system alone. We are looking to fund people working in autonomous systems, robotics, sensing, metrology, data assimilation, probabilistic forecasting, AI, economics, decision science, risk, public policy, product development and systems engineering. We are interested in established teams, emerging organisations and unconventional partnerships that can bring a missing capability into the programme.
We also want to hear from the stakeholders who could help shape, test or ultimately use the system. That includes people working in government and national resilience, financial services, fisheries and marine management, infrastructure, humanitarian response and climate adaptation. We want to see proposals that include co-design partnerships, helping teams define the decisions, lead times, uncertainty thresholds and outputs that would make an early warning genuinely useful.
Greenlandic researchers, institutions and communities are especially important partners. Greenland is not simply a location for collecting data: local expertise and priorities must inform what is observed, how research is conducted, how data are governed, and what useful climate information looks like for the people living closest to these changes. To support this, we have established an Oversight Committee with majority Greenlandic membership to help ensure that research conducted in Greenland champions the Greenland Research Strategy and meets high standards of ethical engagement.
Some applicants may already be part of Forecasting Tipping Points and see an opportunity to extend or combine their work. Others may be entirely new to climate tipping points but have a technology, method or way of thinking that could change what is possible. We will help to build teams that can push beyond the state of the art while working as part of a coordinated programme: teams prepared to share interfaces, test their work against neighbouring modules, and build towards a collective demonstration rather than an isolated result.
The aim of Phase 2 is to build on the work and expertise of our current Creators, while opening the programme to the new capabilities, organisations and collaborations needed to complete the system.
What could this make possible?
Our ambition is to demonstrate an initial end-to-end early warning product against a real user decision in late 2027, while laying the foundations for the partnerships, governance and institutional home needed to sustain the capability beyond ARIA. This is not about claiming that we can predict an exact tipping date. It is about discovering whether trusted observations and forecasts can tell us when a system is changing, when the risks are becoming materially different, and when waiting becomes more dangerous than acting.
Even if a reliable tipping point warning proves impossible, a sustained and adaptive North Atlantic observing and forecasting system would leave a valuable legacy: better information about a rapidly changing ocean, stronger climate preparedness, and new tools for decisions that already need to be made.
That is the challenge for Phase 2 – and the opportunity for our Forecasting Tipping Points Creators, current and future.
Get involved
Our second funding call is planned for 18 September 2026. If your work could contribute to an adaptive ocean observing network, decision-relevant forecasting, or the translation of climate evidence into action, now is the time to begin developing ideas and forming partnerships.
We also want to hear from prospective users, international partners and potential co-funders. Building a capability that endures beyond ARIA will require partnerships with governments, existing ocean-observing programmes, philanthropic and commercial funders, and organisations that could help operate or sustain the future system.
Sign up for ARIA updates to hear when the call opens.