$900,000 Grant to Build AI-Powered Sea-Level Forecasts for Gulf Coast Communities

The Gulf of Mexico is not rising uniformly. Sea-level change in this region reflects a tangle of forces - ocean warming that expands seawater volume, shifts in ocean circulation, and the slow subsidence of land along a coast built on sedimentary deposits that compact and settle over time. Communities in rural Louisiana, Mississippi, Alabama, and Florida face combinations of these factors that differ from their neighbors a few miles away, and they often lack the data infrastructure to understand what those differences mean for their flood risk in 2040 or 2060.

Florida Atlantic University's Harbor Branch Oceanographic Institute has received a $900,000, four-year grant from the Gulf Research Program of the National Academies of Sciences, Engineering, and Medicine to address this problem directly. The project will produce high-resolution, scenario-based sea-level projections for the Gulf region, tailored to local conditions and translated into decision-support tools that communities can actually use.

Why Standard Sea-Level Projections Fall Short

Global and regional sea-level projections - the kind produced by IPCC working groups or national oceanographic agencies - operate at spatial scales that smooth over the variability that matters most for local planning. A projection that estimates a regional average rise of 0.5 meters by 2100 provides limited guidance to a coastal county trying to decide whether to elevate its water treatment plant or rebuild its bridge approaches.

Local projections require combining global climate forcing with regional ocean dynamics, atmospheric patterns, and - critically - vertical land motion. Land subsidence, which occurs when sediments compact, groundwater is extracted, or tectonic forces operate, effectively adds to relative sea-level rise from the community's perspective even when absolute ocean levels remain stable. In parts of the Gulf Coast, subsidence rates exceed the rate of ocean-level rise, making it the dominant component of local relative sea-level change.

The Technical Approach

The FAU team, led by principal investigator Laurent Cherubin, a research professor at Harbor Branch, will integrate multiple modeling frameworks. Community Earth System Model high-resolution simulations will provide the large-scale climate forcing under different emissions scenarios. These will feed into nested regional models - the Semi-implicit Cross-scale Hydroscience Integrated System and Finite-Volume Community Ocean models - that can resolve local coastal geometry at the scales relevant for infrastructure planning.

Vertical land motion data from GPS and satellite altimetry will be incorporated to generate precise, location-specific forecasts that account for subsidence. The team will then apply a graph-based probabilistic machine learning framework to analyze multiple predictors simultaneously - ocean heat content, atmospheric indices, dynamic sea-level trends - to forecast extreme sea-level events and regional variability under conditions of deep uncertainty.

Deep uncertainty is the key phrase here. Projections of future sea-level are inherently probabilistic, and the scenarios that matter most for planning - high-impact, low-probability events - are precisely those that conventional deterministic models handle least well. Probabilistic machine learning approaches can quantify the range of plausible outcomes and communicate that uncertainty in forms that planners can incorporate into risk assessments.

"Being selected for this project is an incredible honor and a great opportunity," said Cherubin. "Our team is committed to advancing the scientific understanding of regional sea-level rise, while translating that knowledge into practical tools that communities can use to plan for the future."

The Community Engagement Component

The project includes a substantive community engagement component, partnering with Florida Sea Grant to work with four Gulf Coast communities - with emphasis on rural and unincorporated areas that lack the municipal resources larger coastal cities use for adaptation planning. The team will conduct outreach programs to build public awareness of sea-level rise scenarios, deploy water level sensors to fill data gaps in undermonitored areas, and co-develop an AI-based platform tailored to local decision-making needs.

The distinction between producing scientific projections and producing useful decision-support tools is one the team takes seriously. High-resolution model outputs that remain in academic papers do not help a county commission deciding whether to permit development in a flood-prone area. The platform aims to bridge that gap with scenario-specific outputs and guidance for planning and emergency response.

"Reliable, actionable and regionally focused sea-level information is a cornerstone of resilience planning in the Gulf," said Michael Feldman, program director of the Gulf Research Program's Gulf Environmental Protection and Stewardship Unit. "The research supported through these awards will provide planners, agencies and communities with the science they need to make forward-looking decisions."

The Funding Source

The Gulf Research Program is an independent science-based program of the National Academies, established in 2013 as part of legal settlements following the 2010 Deepwater Horizon disaster. Its mission focuses on offshore energy safety, environmental health, and the well-being of Gulf region communities. The program has funded a range of environmental monitoring, community health, and resilience research projects across the region.

Co-investigators on the FAU project are Xingquan Zhu from FAU's College of Engineering and Computer Science; Robert Burgman from Florida International University; and Anna Braswell from the University of Florida's School of Forest, Fisheries, and Geomatics Sciences.