arXiv subjects

Omar Sallam

Publications and source records attributed to Omar Sallam.

4 recordsLinked to original sources

arXiv2026

Toward AI-Driven Digital Twins for Metropolitan Floods: A Conditional Latent Dynamics Network Surrogate of the Shallow Water Equations

AI-driven flood digital twins demand fast hydrodynamic surrogates for ensemble forecasting and observation assimilation. Yet even GPU-accelerated two-dimensional shallow water equation (SWE) solvers still require $\sim 55$ minutes per $96$-hour run on a $\sim 4.2$-million-active-cell metropolitan basin (the Des~Plaines River basin at $30\,\mathrm{m}$ resolution), making such workloads prohibitive at native resolution. We present the Conditional Latent Dynamics Network (CLDNet): a low-dimensional latent neural ODE driven by rainfall, paired with a coordinate-based decoder conditioned on static terrain (elevation, slope, Manning roughness) that reconstructs depth and discharge at arbitrary query points. Pointwise decoding decouples memory from grid size and handles irregular watersheds natively, enabling metropolitan-scale training on a single compute node and direct queries at exact gauge coordinates without raster snapping. We evaluate CLDNet on a synthetic $250{,}000$-cell Texas benchmark and on a new Des~Plaines case study of $114$ real-rainfall Stage~IV storms whose reference simulator we validate against United States Geological Survey (USGS) gauges at the April~2013 flood-of-record (Nash--Sutcliffe efficiency $0.57$--$0.94$ on mean-recentered water-surface elevation). CLDNet roughly halves the relative root-mean-squared error of an unconditional baseline, outperforms regular-grid VAE--ConvLSTM and FNO baselines on the Texas benchmark (both presuppose a Cartesian grid and do not apply to the irregular Des~Plaines watershed), reaches a critical success index of $\approx 86\%$ at the $0.5\,\mathrm{m}$ inundation threshold, and produces a full $96$-hour basin-wide forecast in $\sim 29$ seconds -- a $\sim 115\times$ speedup.

cs.LG↗

arXiv2026

Conditional diffusion denoising probabilistic model for super-resolution of atmospheric boundary layer large eddy simulation

Climate change necessitates rapid expansion of renewable energy, with wind energy offering a scalable and low-impact solution. However, accurate prediction of wind loads and power generation remains challenging due to uncertainties in wind shear and turbulence stresses under atmospheric boundary layer (ABL) conditions. High-fidelity Large Eddy Simulations (LES) are typically used to reduce these uncertainties but are computationally expensive and impractical for large-scale or real-time applications. This work addresses this limitation using generative AI, specifically Conditional Denoising Diffusion Probabilistic Models, to reconstruct high-resolution turbulent flow fields from coarse inputs. A high-fidelity dataset is generated using a parallel high-order finite-difference solver across varying geostrophic wind speeds, surface roughness conditions aligned with IEC wind classes, and multiple grid resolutions. The diffusion model is trained for super-resolution across different scale factors and evaluated under interpolation and extrapolation scenarios. Results show accurate recovery of fine-scale turbulent structures, Reynolds stresses, and statistical properties in interpolation cases, indicating strong physical consistency within the training domain. However, extrapolation to higher wind speeds leads to increased noise and overprediction of turbulent stresses, highlighting limitations in generalization. Overall, the study demonstrates that physics-informed generative models can significantly reduce computational cost while maintaining acceptable accuracy, enabling faster and more reliable turbulent inflow characterization for wind energy applications.

physics.flu-dyn↗

arXiv2024

Inference of water waves surface elevation from horizontal velocity components using physics informed neural networks (PINN)

In this paper, a mathematical model is presented to infer the wave free surface elevation from the horizontal velocity components using Physics Informed Neural Network (PINN). PINN is a deep learning framework to solve forward and inverse Ordinary/Partial Differential Equations (ODEs/PDEs). The model is verified by measuring a numerically generated Kelvin waves downstream of a KRISO Container Ship (KCS). The KCS Kelvin waves are generated using two phase Volume of Fluid (VoF) Computational Fluid Dynamics (CFD) simulation with OpenFOAM. In addition, the paper presented the use of the Fourier Features decomposition of the Neural Network inputs to avoid the spectral bias phenomena; Spectral bias is the tendency of Neural Network to converge towards the low frequency solution faster than the high frequency one. Fourier Features decomposition layer showed an improvement for the model learning, as the model was able to learn the high and low frequency components simultaneously.

physics.flu-dyn↗

arXiv2024

Dual scale Residual-Network for turbulent flow sub grid scale resolving: A prior analysis

This paper introduces generative Residual Networks (ResNet) as a surrogate Machine Learning (ML) tool for Large Eddy Simulation (LES) Sub Grid Scale (SGS) resolving. The study investigates the impact of incorporating Dual Scale Residual Blocks (DS-RB) within the ResNet architecture. Two LES SGS resolving models are proposed and tested for prior analysis test cases: a super-resolution model (SR-ResNet) and a SGS stress tensor inference model (SGS-ResNet). The SR-ResNet model task is to upscale LES solutions from coarse to finer grids by inferring unresolved SGS velocity fluctuations, exhibiting success in preserving high-frequency velocity fluctuation information, and aligning with higher-resolution LES solutions' energy spectrum. Furthermore, employing DS-RB enhances prediction accuracy and precision of high-frequency velocity fields compared to Single Scale Residual Blocks (SS-RB), evident in both spatial and spectral domains. The SR-ResNet model is tested and trained on filtered/downsampled 2-D LES planar jet injection problems at two Reynolds numbers, two jet configurations, and two upscale ratios. In the case of SGS stress tensor inference, both SS-RB and DS-RB exhibit higher prediction accuracy over the Smagorinsky model with reference to the true DNS SGS stress tensor, with DS-RB-based SGS-ResNet showing stronger statistical alignment with DNS data. The SGS-ResNet model is tested on a filtered/downsampled 2-D DNS isotropic homogenous decay turbulence problem. The adoption of DS-RB incurs notable increases in network size, training time, and forward inference time, with the network size expanding by over tenfold, and training and forward inference times increasing by approximately 0.5 and 3 times, respectively.

physics.flu-dyn↗