Defining nationally consistent coastal flood severity thresholds

Objective:

This project aims to develop a machine learning-based framework that generates spatially consistent coastal flood severity thresholds every 10 km along the U.S. coastline to enhance flood risk communication, planning, and resilience.

Approach:

To overcome the limitations of point-based flood severity thresholds, we are developing a machine learning (ML) framework that learns from both physical data (e.g., sea level rise, elevation, and tides) and observed impacts (e.g., NWS advisories) to generate consistent minor, moderate, and major flood thresholds across the U.S. coastlines. Our model considers a range of features and identifies patterns that define flood severity in a locally relevant way.
The ML model is trained and validated using existing NOAA tide gauge records, historical advisories, and flood impacts. Once trained, the model will extrapolate thresholds to ungauged locations at a 10-km resolution. The framework is being co-developed and evaluated in collaboration with NOAA-NOS scientists and validated using local datasets and soon based on the feedback from regional agencies (e.g., South Florida Water Management District). Inter-agency engagement with FEMA and EPA further ensures the product meets operational needs and complements national hazard planning priorities.

Impact:

Abstract:

Project Introduction:
This project develops a machine learning–based framework to generate spatially consistent coastal flood severity thresholds—minor, moderate, and major—every 10 km along the entire U.S. coastline. These thresholds integrate physical drivers (e.g., sea level, elevation, and tides) with observed impacts (e.g., NWS advisories, and reports) to improve the accuracy and clarity of flood risk communication and decision-making.

Context:
Currently, NOAA’s National Weather Service provides flood severity thresholds only at a limited number of gauged locations. These thresholds are inconsistent across regions and often fail to reflect real-world flood impacts. As sea levels rise and development in coastal zones accelerates, the inability to capture consistent, localized flood thresholds hampers effective public communication, risk mitigation and planning. Our project addresses this challenge by creating a nationally consistent, data-informed system to monitor and contextualize flood severity at a much finer spatial resolution.

Research Concept and Significance:
We identified the need for consistent, spatially detailed flood severity thresholds that go beyond the limitations of point-based tide gauge data. Our machine learning approach uncovers relationships between physical drivers and historical flood impacts to define thresholds in ways that are both locally relevant and nationally standardized.

Product and Intended Advancements:
We aim to deliver a robust, validated flood threshold mapping framework and associated datasets designed for direct integration into NOAA operations and community-level planning.

What is being done today:

- Flood thresholds are defined only at limited tide gauge sites.
- Thresholds may not correspond to observed impacts, leading to public confusion.
- No consistent national framework exists for defining or applying flood severity categories.

How this project advances the field:

- Develops a machine learning model to infer flood thresholds at 10-km resolution along the entire U.S. coastline.
- Integrates a wide array of physical and impact-based data sources.
- Provides an operational tool co-developed with NOAA NOS and validated through interagency and local partnerships.
- Supports dynamic updates and future applications in climate change adaptation.

Impact: This project will fundamentally enhance the nation’s coastal flood risk communication and planning capabilities by:
- Enabling more accurate and location-specific flood warnings, reducing false alarms and overlooked risks.
- Informing efficient flood mitigation strategies by providing data for underserved and ungauged areas.
- Standardizing flood severity definitions across agencies and regions, improving national coordination.
- Strengthening interagency collaboration, already evidenced through active engagement with NOAA NOS, FEMA, and EPA, and extending these benefits to regional planners and community decision-makers.

Ultimately, this project will empower coastal communities and federal partners alike with the tools to better anticipate, communicate, and respond to flooding threats in a changing climate.