rSHUD: An R Package Facilitating Quick and Reproducible Hydrologic Deployment Worldwide.

Abstract:

Modern hydrologists leverage a number of readily available, online databases describing hydrologically relevant quantities. These data resources can vary substantially, and their complexity and heterogeneity impact interoperability and reproducibility in scientific research. This work focuses on streamlining the necessary data processing steps in support of rapid hydrological model implementation anywhere in the world and highlights the best practices in modeling to tackle the challenge of reproducibility in a hydrological context.

The Simulator for Hydrologic Unstructured Domains (SHUD) is an integrated hydrological surface-surface model with a finite volume method, that supports modeling from the hillslope to the continent-scale. Paired with the SHUD model, the rSHUD is an open-source R package for model pre- and post-processing, including spatial data processing, unstructured domain discretization, model input preparation, automated calibration, hydrological analysis, and 2D/3D visualization. With support combined with open-access global and national datasets, rSHUD enabled hydrological simulations, calibration, and visualization for three representative watersheds in China, the US, and Ethiopia. Thus the SHUD modeling system offers a new paradigm for reproducible watershed modeling, thanks to the automatic model deployment and extensibility to other modeling systems.