the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Jul 2026
GeoFlood (v1.0.0): Computational model for overland flooding
Abstract. This paper presents GeoFlood, a new open-source software package for solving the shallow-water equations (SWE) on a quadtree hierarchy of mapped, logically Cartesian grids managed by the parallel, adaptive library ForestClaw (Calhoun and Burstedde, 2017). The GeoFlood model is validated using standard benchmark tests from Neelz and Pender (2013) as well as the historical Malpasset dam failure. The benchmark test results are compared against those obtained from GeoClaw (Clawpack Development Team, 2020) and the software package HEC-RAS (Hydraulic Engineering Center - River Analysis System, Army Corps of Engineers) (Brunner, 2018). The Malpasset outburst flood results are compared with those presented in George (2011) (obtained from the GeoClaw software), model results from Hervouet and Petitjean (1999), and empirical data. The comparisons validate GeoFlood's capabilities for idealized benchmarks compared to other commonly used models as well as its ability to efficiently simulate highly dynamic floods in complex terrain, consistent with historical field data. Because it is massively parallel and scalable, GeoFlood may be a valuable tool for efficiently computing large-scale flooding problems at very high resolutions.
Status: final response (author comments only)
- EC1: 'Comment on egusphere-2026-3967', Donghui Xu, 24 Jul 2026
-
RC1: 'Comment on egusphere-2026-3967', Anonymous Referee #1, 27 Jul 2026
This manuscript presents the GeoFlood model, which may be useful for flood-inundation simulations over complex terrain. However, substantial revision is required before the manuscript can be considered for publication in Geoscientific Model Development.
(1) Novelty: The main concern is that the scientific and software novelty of GeoFlood is not yet sufficiently established. GeoFlood is built upon mature existing components, including GeoClaw solvers, ForestClaw, and p4est, and the manuscript describes the model as an integration of these libraries with additional AMR and visualization routines. It is unclear whether the principal contribution is a new numerical method, a new scalable AMR implementation, or primarily an integration of existing packages.
In the Introduction, the authors wrote "developing numerical models based on SWE remains a challenging problem with an active research community". However, there are tons of SWE-based flood models, such as LISFLOOD-FP, Telemac-2D, SERGHEI-SWE, Triton, etc. The authors should explicitly identify GeoFlood’s core advantage over existing flood models. This comparison should address the governing equations, numerical schemes, wetting-and-drying treatment, mesh structure and refinement strategy, parallelization approach, computational efficiency, and supported physical processes. If AMR is the principal innovation, the manuscript should demonstrate quantitatively why it offers a meaningful benefit over other established approaches.
(2) Parallelization: The manuscript emphasizes massively parallel and scalable flood simulation, but the reported performance assessment is CPU-based and limited to a single 48-core node. Given that many modern flood models increasingly exploit GPU acceleration, the computational competitiveness of GeoFlood is not yet demonstrated convincingly.
GPU support is not necessarily required for publication. However, if GPU acceleration is outside the scope of the present model, the authors should position GeoFlood as a CPU/MPI-based AMR framework rather than as a generally high-performance or state-of-the-art parallel flood model. The manuscript should also discuss the implications of this design choice for large-scale applications.
Furthermore, although Fig. 16 indicates higher parallel efficiency for GeoFlood than for GeoClaw, the absolute wall-clock-time reduction appears limited. A more detailed performance analysis would be valuable, including the computational costs of numerical updates, AMR regridding, inter-process communication, etc. Such profiling would clarify the principal bottlenecks and indicate which parts of the code are most difficult to parallelize or optimize.
(3) Test Cases: The four test cases primarily examine dry-bed inundation driven by prescribed inflow or dam-break release, wetting and drying, topographic effects, and AMR behavior. While these are relevant tests, they represent similar hydraulic scenarios and do not demonstrate a comprehensive flood-modeling framework.
The manuscript does not demonstrate rainfall forcing, evaporation, infiltration, stage boundary conditions, rainfall-runoff generation, urban features, or other potentially relevant hydrological and hydraulic processes. The authors should therefore either add benchmark or application cases for each additional capability that GeoFlood claims to support, or clearly define GeoFlood as a model for prescribed-flow and dam-break-type inundation, rather than implying that it is a general-purpose flood model.
(4) Quantitative Analysis: The comparisons with GeoClaw and HEC-RAS are mainly visual or qualitative, with results often described as “consistent” or “comparable.” More quantitative evaluation is needed. For each test case, the authors should report appropriate skill metrics, such as water-level and velocity errors, inundation-area agreement, mass-balance error, and computational cost. If AMR is the principal contribution of this study, a quantitative AMR convergence and efficiency analysis is also essential.
Citation: https://doi.org/10.5194/egusphere-2026-3967-RC1 -
RC2: 'Comment on egusphere-2026-3967', Anonymous Referee #2, 27 Aug 2026
The manuscript presents the GeoFlood model. 3 benchmark tests and the realistic case Malpasset are considered. The authors have made a good effort in developing the GeoFlood model. The use of shallow water equations in conjunction with adaptive mesh refinement (AMR) is potentially useful for flood inundation applications. However, the manuscript requires significant revision before it can be considered for publication in Geoscientific Model Development.
1) Regarding introduction and novelty: Firstly, the introduction would benefit from better communication of the motivation for developing GeoFlood (the conclusion does this to an extent), particularly the specific research gap in existing flood models that GeoFlood is intended to address. In its current form, the manuscript presents GeoFlood simultaneously as SWE solver (from Clawpack, GeoClaw), an AMR approach (from ForestClaw), a visualization tool, which makes the main contribution unclear. Clearly defining the gap that motivated GeoFlood and explicitly stating the specific objectives of the study would sharpen the manuscript, better articulate the manuscripts principal contribution and assist in giving the model a clearer identity. Coming to novelty, the authors state “In this paper, we present GeoFlood, a new computational model”, the claim for new model needs to be supported with additional details. If the novelty is associated with AMR or computational efficiency, this should be supported by quantitative evidence (errors, compute time).
Secondly, the statement that “The potential for future devastating floods caused by the failure of still-operational dams remains” should be supported with additional literature beyond the Mosul dam example. Recent studies on aging dam infrastructure and changing hydrologic risks would strengthen this motivation. Next, the statement that "developing numerical models based on SWE remains a challenging problem with an active research community" should be expanded. Numerous established SWE flood models exist, and authors should briefly discuss other models and clarify how GeoFlood differs from them in terms of numerical discretization, solvers, AMR in the context of the problems being solved. Essentially, the authors should present an argument for where and why GeoFlood is a suitable choice over some of the existing models for flood inundation.
2) Benchmark test cases: The manuscript would benefit from a numerical verification and convergence study (error vs grid-resolution dx) against a known solution. I recommend adding at least one analytical benchmark and one-two other problems (that widen GeoFlood applicability), run at multiple resolutions. Accordingly, the errors should be reported quantitatively for all test cases. In addition, for one simple problem and for one realistic problem, comparing errors in depth against computational time, with and without AMR (for a sequence of grid-resolutions; the finest uniform-grid maybe a reference solution if analytical solution does not exist), would provide a meaningful assessment of the accuracy-efficiency trade-off.
3) AMR refinement criteria: Section 4.1 presents several refinement criteria for GeoFlood, and the authors have made a good effort to identify physically relevant AMR indicators. The authors may optionally consider evaluating Froude Number, or a similar flow-based refinement criterion, that compactly captures the effects of both flow velocity and water depth together.
4) Schematic Figure: For readers who may be less familiar with AMR, the manuscript would benefit from improving Figure 1 to also have simple schematic figure illustrating the minimum and maximum refinement levels with clear labels (say for a simple 2D problem). The figure could show how the mesh changes across refinement levels and how each level corresponds to the spatial resolution used in GeoFlood. It could also indicate the theoretical computational savings of AMR relative to a uniformly fine mesh, thereby clarifying the efficiency gained through local refinement and this brings me to a related technical point. The authors should clarify how the CFL-controlled time step varies across AMR levels and whether temporal subcycling is employed (as indicated in ForestClaw), as this directly influences the computational efficiency of their AMR approach.
Citation: https://doi.org/10.5194/egusphere-2026-3967-RC2
Data sets
Datasets used in GeoFlood comparison to other models Brian Kyanjo https://doi.org/10.5281/zenodo.10897305
Model code and software
GeoFlood model Brian Kyanjo and Donna Calhoun https://doi.org/10.5281/zenodo.10929142
User Manual for running all codes used in the paper Brian Kyanjo https://drive.google.com/file/d/1O3QizHHNUrOUjw6Uw-G_2tLBPcZgT-PP/view?usp=sharing
Viewed
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 183 | 0 | 5 | 188 | 0 | 0 |
- HTML: 183
- PDF: 0
- XML: 5
- Total: 188
- BibTeX: 0
- EndNote: 0
Viewed (geographical distribution)
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
I encourage the authors to consider the comments from the previous review posted on NHESS Open Discuss: https://doi.org/10.5194/egusphere-2025-2173-RC1.