Welcome to the repository for "Uncertainty in Estimating the Relative Change of Design Floods under Climate Change: a Stylized Experiment with Process-Based, Deep Learning, and Hybrid Models".
This study explores uncertainty in estimating changes in design floods using six hydrological models - including process-based, deep learning, and hybrid approaches - across 30 basins in Massachusetts, USA. We evaluate how model structure, input error, and climate scenarios affect relative change estimates to support reliable hydrological projections for long-term planning.
This paper has been published in Journal of Hydrology.
Read the paper here: https://www.sciencedirect.com/science/article/pii/S0022169425017676?via%3Dihub#b0225
Cite: Poudel, S., Najibi, N., & Steinschneider, S. (2025). Uncertainty in estimating the relative change of design floods under climate change: A stylized experiment with process-based, deep learning, and hybrid models. Journal of Hydrology, 664, 134427. https://doi.org/10.1016/j.jhydrol.2025.134427
This repository is organized into the following sections:
- Contains scripts for data processing, model training and analysis, and result visualization. A brief description of each script is provided in the accompanying README file.
- Contains scripts for sub-experiments conducted in the study. This includes sensitivity analysis on results from the main experiment. A brief description of each sub-experiments is provided in the accompanying README file.