This project investigates ways to reduce the error in projected changes in design flood estimates (e.g., % Change in 50-year flood) under climate change scenarios by leveraging regional pooling strategies. We analyze results from process-based, deep-learning based, and hybrid hydrological models applied across 30 basins in Massachusetts, and compare different pooling techniques to improve the accuracy of flood change estimates.
-
data/
Contains input data, including basin-level covariates and model-estimated changes in 50-year flood estiamte under future climate conditions. -
scripts/
Contains all Python and SLURM scripts to run the full analysis.