Title: PortUrb: A Performance Portable, High-Order, Moist Atmospheric Large Eddy Simulation Model with Variable-Friction Immersed Boundaries

In this paper, the PortUrb code is introduced. This code is designed for moist atmospheric LES with a focus on urban environments, emphasizing portability, performance, accuracy and clarity. It leverages portable C++ libraries and GPU-friendly numerical methods like WENO schemes to ensure high performance and robustness across various hardware platforms. Numerical experiments validate its accuracy and robustness across multiple test cases, including atmospheric boundary layers, urban flow around buildings, and realistic geometries, demonstrating reliable, physically accurate performance without numerical artifacts. The topic is important, and the manuscript is generally well-written and streamlined. The scientific merit of the study deserves publication. Yet, I recommend minor revision of the manuscript before its acceptance. This recommendation is based on the comments and remarks listed below:

• In lines 55–58, the authors note that a series of numerical discretization studies led to their choice of discretization schemes. To enhance the manuscript's clarity and value to readers, it would be helpful if the authors briefly discussed the rationale behind selecting these particular schemes in the introduction. What specific advantages or limitations influenced their decision? Including this context—such as trade-offs in accuracy, stability, or computational cost—would strengthen the motivation and transparency of their methodological choices.

• While the introduction emphasizes the portability of the model, it is unclear how this is demonstrated in the manuscript. Could the authors clarify what specific aspects of the code design or implementation contribute to its portability across architectures? Additionally, it would be helpful to include performance metrics—such as total runtime or computational resources used—for the Neutral Atmospheric and Dry Convective Boundary Layer test cases. This information would give readers a better understanding of the model’s efficiency and scalability.

• It is not clear how geostrophic forcing enters Equation 1. Could the authors explicitly state how this term is incorporated into the governing equations? Additionally, in line 297, the statement that “Equation 18 is applied using the average column” is ambiguous. What does “average column” refer to?

• The manuscript contains several references to figures in other studies when comparing results. For instance, in line 310, the authors state that “these profiles closely match Figure 3 in Sauer and Muñoz-Esparza (2020),” which requires readers to consult external sources while reading. To improve readability and make the manuscript more self-contained, I recommend including the relevant figures from these referenced studies alongside the corresponding figures in this manuscript (e.g., overlaying Figure 3 from Sauer and Muñoz-Esparza (2020) with Figure 3 here). This approach would allow readers to directly assess the comparison without referring to external documents. A similar suggestion applies to Figure 2 and other instances where visual comparisons are made.

• In line 306, the authors mentioned that “the smoothing of the second temperature discontinuity is lower in portUrb than seen in FastEddy”. The author did not provide any evidence to support this statement.

See attached document with my review comments.