Regional Modeling: Improving Estimates of Changes to Florida’s Rainfall

Floridians are familiar with characteristic summer storms and other events delivering heavy rainfall, but will these events become more intense in the future? The unique geography of our state, being a narrow peninsula with an extensive coastline, calls for detailed computer modeling to estimate how rainfall will change under future conditions.  

To that end, the Florida Flood Hub for Applied Research and Innovation has partnered with the Florida Department of Environmental Protection, Florida Department of Transportation, South Florida Water Management District, University of Miami, Florida State University, and the United States Geological Survey to improve estimates of future extreme rainfall for the state. To do so, they are coupling a series of computer models to simulate precipitation over the entire Floridan Aquifer.     

Scientists use global circulation models to simulate the physical processes driving the Earth’s climate, portraying global patterns in temperature, rainfall, ocean currents, and other factors. These global models have a coarse spatial resolution, meaning they trade off local details to allow coverage of a very large area. As a result, local features that affect rainfall in Florida are not captured well.  

Comparison of resolution from (a) a global climate model (CESM2) and (b) the regional climate model (RSM-ROMS) at 10 km. The red outline delineates the boundary of the modeling, which covers the entire Floridan Aquifer. (Adapted from Misra and Jayasankar, 2025)

One local feature that is not captured well in global models is the Loop Current, a strong ocean current that transports warm waters from the Caribbean through the Yucatan Channel and “loops” around Florida’s peninsula. The heat in this current can drive sea-breeze thunderstorms in the wet season. By failing to account for the Loop Current, global circulation models may miss a large part of the extreme rainfall that occurs in Florida. 

To obtain future rainfall estimates that account for local-scale features, outputs of global models are downscaled to a finer resolution. There are two main approaches to downscaling: statistical and dynamical. 

Statistical downscaling relies on mathematical relationships between outputs from global models and local observations within a historical period. These relationships are then applied to outputs from modeling that estimates future rainfall to yield more localized estimates.  

However, this method relies on the assumption that historical relationships remain valid in the future, which is deemed less likely given recent observations. In addition, statistical downscaling does not allow estimated changes to be linked to physical drivers – the downscaling process has decoupled the results from the dynamics of the drivers. 

The regional climate models being developed by the partnership use an alternative approach known as dynamical downscaling, which continues to rely on equations that describe the physical interactions between factors that drive rainfall, rather than switching to statistical relationships. In this approach, the outputs of a model with lower spatial resolution get fed into another model with higher resolution. Ultimately, the final outputs apply locally and remain connected to key dynamics. Importantly, dynamical downscaling allows us to understand both the “what” and the “how” underpinning changes in rainfall. 

“We are reaching unprecedented resolution in simulating rainfall over Florida while retaining the storyline of Florida’s hydroclimate” said Vasu Misra, professor of meteorology at FSU serving as one of the principal investigators for the project. 

The participating researchers will use their models to simulate three 10-year scenarios: one for conditions during 2011–2020, and two others based on warmer conditions (2°C and 3°C increase in global air temperature). The output will be an ensemble of simulations with spatial resolution of four km2 or as detailed as one km2 for a few extreme events. Comparing results from simulations that include a warmer future to simulations of the 2011– 2020 baseline will highlight key changes in rainfall. 

The changes that are identified will help scientists, practitioners, and policymakers understand future patterns of precipitation and the associated drivers, which is essential information for communities as they prepare for extreme rainfall in the future.