Reserve Data Matches up with Machine Learning to Track Storm Recovery

Before and after Hurricane Irene hit the Everglades.

After a hurricane hits, some damage is easy to spot—flooded neighborhoods, downed trees, widespread power outages. But some impacts are more subtle, including changes in the water quality of estuaries, which protect some of the country’s most economically important habitats.

Now, thanks to long-term data from the National Estuarine Research Reserve System-Wide Monitoring Program (NERRS SWMP), researchers at the University of Alabama in Huntsville have built a machine learning model that can help communities understand how estuary water quality is affected by cyclones and estimate how long it will take to recover. 

“Practically speaking, this model can act like a diagnostic tool for decision makers,” says Abdullahi Salman, the associate professor of Civil & Environmental Engineering who worked with colleagues Muhammad Ahsan Ibrar and Muhammad Usama on the model. “After a storm, they can use it to efficiently measure the level of disturbance and track water quality recovery over time, crucial information for activities like planning restoration efforts and reopening fisheries.”

Estuaries are dynamic, complex systems. Salinity, oxygen, temperature, and nutrient levels shift constantly, and often unpredictably. That complexity has made them tough to model using traditional statistical methods. But SWMP’s decades-long dataset—collected at 30 Reserves using consistent methods—gave the research team the detailed baseline they needed to train a more accurate, scalable model.

“Reserve data was absolutely key to building our model,” says Salman. “The NERRS program provided a rare dataset—high-frequency, long-term water quality data with readings taken every 15 minutes over many years. This level of detail enabled us to train a model capable of detecting even small disruptions in estuarine conditions.”

“The data collected through NERRS SWMP is an invaluable resource for scientists interested in developing tools to support coastal communities because it is quality data that can be trusted,” says Rebecca Roth, executive director of the National Estuarine Research Reserve Association (NERRA). “Decades of federal and state investment in this dataset has yielded a mission-ready resource that scientists across the country are using to deliver real-time, actionable tools and information for the coasts.”

When water quality is disrupted, it can impact fish health, reduce catches, and affect the economy. By quantifying how long those disruptions last, the model offers valuable insight for managing fisheries, protecting habitats, and planning economic recovery.

Although developed with U.S. coastal data, the model can be adapted to estuaries around the world—or even freshwater systems like the Great Lakes—where comparable monitoring data exists.