the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Jul 2026
Implementation of a resolved building capability in the Energy Research and Forecasting model for accelerated high-fidelity urban-scale simulation
Abstract. Accurate predictions for wind and particle dispersion must be obtained quickly to inform emergency response to atmospheric releases of hazardous material in urban areas. In this study, we demonstrate and validate a newly implemented resolved building capability added to the exascale Energy Research and Forecasting (ERF) model. To represent buildings, an immersed forcing method is employed, which enforces near-zero velocities within buildings. We modify the immersed forcing method to implement a wall-model where we enforce a calculated velocity for building walls determined using the log law. The immersed forcing method is evaluated by simulating cases from the Joint Urban 2003 Oklahoma City tracer release experiment, analyzing cases with both stable and unstable thermodynamic conditions. For the stable case, we use adaptive mesh refinement for regions above a certain scalar concentration. Additionally, we conduct two sensitivity studies: one on building wall roughness length, finding that 0.1 m results in the most accurate scalar concentration predictions, and another on model configurations, where computational cost decreases by an order of magnitude when either not refining the grid or when using anelastic mode compared to fully compressible mode, with model skill only decreasing slightly. For the convective case, model skill is not as high as the stable case largely due to underprediction of scalar concentration at stations to the west and upwind of the release site. Ultimately, model skill for both cases are still among the highest in the published literature, demonstrating the efficacy of our method for accelerated high-fidelity urban atmospheric modeling applications.
- Preprint
(16303 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-4030', Anonymous Referee #1, 31 Aug 2026
-
RC2: 'Comment on egusphere-2026-4030', C. GarcÃa-Sánchez, 14 Sep 2026
The paper includes high-resolution urban flow and dispersion simulations for the Oklahoma Joint Urban 2003 case. The researchers used the exascale model ERF to perform high-fidelity simulations under diverse stratification conditions (stable and unstable) and compared dispersion and winds during those periods. The boundary conditions for the numerical experiments were fixed, meaning that only turbulence varied at the boundary, while wind direction was conserved along the simulation. In the paper, an innovation is the implementation of the wall model near the buildings, which improves the model's predictive skill. It was a really nice read, and the paper presents further development in the field of atmospheric flows, so I would support its publication after a few minor comments/questions.Â
General comments:
- Although the model shows one of the highest skill levels in predicting these specific IOPs and conditions, it would be helpful for the reader to have a model limitations section, for example, the minimum number of GPUs required to run efficiently, assumptions regarding subgrid modeling, and the effects of the immersed boundary method.Â
- While reading the article, I kept thinking what makes ERF different from FastEddy in essence? Could the authors clarify this?
- I prefer lighter backgrounds in dispersion plots (fig.9 and 17), as darker backgrounds remove the details of the plume, which look very beautiful on a white background. No need to change if the authors prefer this background.
Abstract: The abstract could be a bit more specific about the innovation in the paper; the content is there, but it could be made more explicit, for example, by mentioning the wall model implementation, which seems to be one of the innovations here.
Introduction:
- "Accurate simulation is required to provide guidance to officials to ensure public safety" --> I would argue that, for a daily air quality prediction, it has been shown that lower fidelity models (not LES) can perform really well to guide officials regarding policy. The strength of the ERF model lies in fast and dangerous releases, where, for time and safety reasons, it might be better to model plume dispersion accurately, especially given current access to GPUs. I would recommend reflecting on the different dispersion problems, those that refer to the reduction of years of life, and those that pose an imminent threat to health.Â
- "Representing urban environments using an exact method such as solid boundaries is generally superfluous as there are a number of subgrid-scale factors that need to be parameterized, and for which limited data exist" --> I don't fully understand this sentence. What subgrid-scale factors need to be parameterized? Why does limited data exist? What about resolving buildings explicitly? This reads as if the immersed boundary method might be better than an explicit mesh when modeling around buildings, but I believe there is currently no such support in the literature, right? The immersed boundary method is indeed the way to go if you are including weather predictions at the boundary, but many other authors have published using explicit surfaces in LES, which is especially important for unsteady problems such as wind loading. Could the authors clarify and include references for their reasoning?
Methods:Â
- "The increase in performance for using anelastic mode compared to fully compressible model will vary and need to be optimized from problem to problem" --> I would expect this to be agnostic about the problem, could the authors elaborate why is this problem dependent?
- Equation (2)--> 0.5 coefficient, is it related to the central point in the first cell? I tried to find it in the text, but couldn't figure it out.
- Equation (3)--> 1.5 coefficient, same as the previous question.
- How were the ranges for z0w chosen? Does this have any relationship with the geometry itself? I could not find it in the article; maybe I missed it?
- "These characteristics will vary from building to building" --> will this mean that we need a z0 per building? How could we determine that?
Modeling setup:Â
- "We also use a short 50m sponge layer to dampen mechanical turbulence at the outflow" --> Is this a feature from ERF, or it was included for this specific numerical experiment? and could others reproduce this with the information provided in the article?
- "with uniform grid spacing delta=5m" --> also in the vertical direction? Should it not be good to have a higher resolution vertically? Why the authors chose this?
- "An outer domain with doubly periodic boundary lateral boundary conditions is used to spin up convectively generated turbulence" --> Does this mean that the velocity at the inflow varies along the 1800s? Or is the mean profile the same?
- "We also buffer the tagging regions in the spanwise (x) dimension by 250m and in the streamwise (y) dimension by 750m" --> Is there a physical reasoning for these dimensions? Was it based on trial and error? How would other researchers attempt to do this domain definition a priori using this publication?
- "We also specify a relatively low grid efficiency for the tagging algorithm" --> what does this practically mean?Â
- Fig.6: I was surprised to see that the start of the simulation uses a fixed velocity profile. Wouldn't it make sense to start from a logarithmic profile instead? Would it change the run time? Also, please include the shaded part in the legend for clarity.Â
Results and discussion:Â
- Fig. 10: I think adding the roughness value at the top in a box might make reading the plots nicer.Â
- page 23: "slightly lower for the anelastic" --> it is difficult to say what is slightly; could the authors remove that adjective, just lower is enough.Â
- "While accuracy is slightly lower in terms of SF6 concentration, the decrease in computational cost makes the anelastic configuration a promising option for emergency response" --> I think it might be relevant here to double-check if anelastic is consistently overpredicting or underpredicting concentration? Could the authors comment?
- "Additionally the absense of thermal wall forcing likely leads..." --> at this point of the article it is unclear why this is mentioned, maybe I missed it in the model set-up, but maybe this should be highlighted in that section? Why is this forcing meaning and why it was not included?
Conclusions:Â
- "fastest urban modeling tools that currently exist" --> I agree that the tool is fast, but what does this "fast" refer to? Does this include the pre-processing and post-processing time? With such large, complex urban databases, those aren't negligible anymore. Are we only talking about computing time? Could the authors comment on this? if pre-post were not included here, could the authors comment on the time spent to make those work? Was it more than the run time?
Â
Â
Â
Â
Â
Citation: https://doi.org/10.5194/egusphere-2026-4030-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 242 | 111 | 39 | 392 | 25 | 31 |
- HTML: 242
- PDF: 111
- XML: 39
- Total: 392
- BibTeX: 25
- EndNote: 31
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The paper presents and validates a newly implemented resolved building feature added to the exascale Energy Research and Forecasting (ERF) model, set up to predict wind and particle dispersion during the Joint Urban 2003 Oklahoma City tracer release experiment. The model examines scenarios under both stable and unstable thermodynamic conditions, focusing on building wall roughness length as a key parameter for near-wall parameterization. Various model configurations are tested, including adaptive mesh refinement and both compressible and anelastic solvers.
The content is relevant to the scope of GMD, and the results merit publication; however, the manuscript needs revisions in several areas because the material is not always sufficiently detailed or accurate.
Attached is the detailed review with specific comments on aspects that require improvements.