OSE-FLOW

Open Science Environment for FLOod and Water hazard modeling.

OSE-FLOW is the Flood Lab's consolidation of the CREST model family into one place, so that the pieces these models share — water balance, routing, hydrodynamics, calibration, differentiable training — can be developed as modular components instead of being re-implemented in each codebase.

Status: archive stage. This repository currently holds verbatim snapshots of the source models, nothing more. No harmonization, refactoring, or shared interface exists yet. Each tree under models/ builds and runs exactly the way its upstream README describes, with its own toolchain and its own conventions.

The models

Directory	What it is	Language / stack
models/EF5	Ensemble Framework For Flash Flood Forecasting. The backbone. A distributed hydrologic modeling framework built for continental-scale flash flood forecasting, carrying three water balance models (CREST, SAC-SMA, hydrophobic) coupled to linear-reservoir or kinematic-wave routing. Created by the HyDROS Lab at the University of Oklahoma.	C++, autotools
models/CREST-iMAP	CREST inundation MApping and Prediction, v2. Couples the CREST water balance to a well-balanced finite-volume hydrodynamic core, differentiable end to end. Produces inundation extent and depth, not just discharge.	Python, PyTorch
models/CREST-VEC	CREST with vector-based routing. EF5/CREST runoff generation coupled to mizuRoute river-network routing, so simulation follows the actual vector river network across scales.	C++ and Fortran
models/dCREST	Differentiable CREST. A distributed, physics-informed CREST for alpine catchments with global transfer learning, targeting cryospheric runoff dynamics. Release code for the WRR paper "Diagnosing Cryospheric Runoff Dynamics."	Python, PyTorch
models/iCRESLIDE	Integrated CREST + SLIDE, v2.0. Couples CREST hydrology to the SLIDE slope-stability model for rainfall-triggered landslide prediction, with a Latin-hypercube ensemble driver for uncertainty and risk analysis. Shipped as MATLAB P-code — see the caveat below.	MATLAB (.p bytecode)

Between them these span the axes the lab works along: process backbone (EF5), spatial routing representation (VEC), hydrodynamic coupling (iMAP), learnable parameterization (dCREST), and multi-hazard coupling (iCRESLIDE).

iCRESLIDE caveat. Unlike the others, iCRESLIDE is distributed as MATLAB P-code — 33 of its 34 files are .p bytecode, which runs under MATLAB but cannot be read, diffed, or modified, and cannot be decompiled back to .m. Only ensemble_simulation.m is readable source. It is archived here as a runnable reference, but it cannot be harmonized or modularized without .m sources from the authors. Its >1 GB test dataset lives on Google Drive upstream and is not mirrored here.

Where the code came from

Every tree under models/ is a snapshot of an upstream repository or release bundle, taken 2026-08-23, with the upstream .git removed. Exact source URLs, commit SHAs, licenses, and what was intentionally left out are recorded in PROVENANCE.md. Read it before treating any of this as canonical — upstream repositories are still the source of truth until harmonization begins.

Licensing

Licenses are per-model and inherited from upstream, not uniform across this repository. EF5 and CREST-VEC are released into the public domain (Unlicense). CREST-iMAP and dCREST carry no declared license upstream; treat them as all-rights-reserved and contact the authors before reuse. See PROVENANCE.md for the per-model table.

Citing

Cite the model you actually used, not this repository:

• EF5 — Flamig, Z. L., Vergara, H., & Gourley, J. J. (2020). The Ensemble Framework For Flash Flood Forecasting (EF5) v1.2. Geoscientific Model Development, 13, 4943–4958. https://doi.org/10.5194/gmd-13-4943-2020
• CREST-VEC — Li, Z., Gao, S., Chen, M., Gourley, J., Mizukami, N., & Hong, Y. (2022). CREST-VEC: a framework towards more accurate and realistic flood simulation across scales. Geoscientific Model Development, 15, 6181–6196. https://doi.org/10.5194/gmd-15-6181-2022
• CREST-iMAP — Li, Z., Chen, M., Gao, S., Luo, X., Gourley, J., Kirstetter, P., Yang, T., Kolar, R., McGovern, A., Wen, Y., Rao, B., Yami, T., & Hong, Y. (2021). CREST-iMAP v1.0: A fully coupled hydrologic-hydraulic modeling framework dedicated to flood inundation mapping and prediction. Environmental Modelling & Software, 141, 105051. https://doi.org/10.1016/j.envsoft.2021.105051 (v2 has no dedicated paper yet — see its README for the second CREST-iMAP reference.)
• iCRESLIDE — see models/iCRESLIDE/README.md and the upstream repository, which do not name a paper; contact the authors.
• dCREST — Diagnosing Cryospheric Runoff Dynamics: A Distributed Differentiable Hydrological Model with Global Transfer Learning. Water Resources Research (see models/dCREST/README.md).

Roadmap

1. Archive — mirror the source models verbatim. (current stage)
2. Harmonize — identify shared components across the readable trees and settle on common I/O, configuration, and grid conventions.
3. Modularize — factor the shared components out so each model composes them rather than carrying its own copy.

---

Maintained by the Flood Lab · https://hydrors.us/