Leveraging Statistical Terrain Surrogates for Rapid Wildfire Management Decisions in Complex Terrains

Effective wildfire management critically depends on the ability to rapidly generate accurate model predictions to inform operational decisions. However, physics-based wildfire propagation solvers are accurate, but computationally expensive and too slow for real-time applications during active wildfire events. To address this challenge, we introduce a surrogate modeling approach based on Statistical Shape Modeling (SSM) for efficient terrain analysis and reconstruction using Digital Elevation Model (DEM) data. Focusing on complex terrains in southwestern New Mexico, terrain features were extracted through percentile-based filtering, followed by preprocessing via morphological dilation, centering, and rotational alignment. Principal Component Analysis (PCA) provided a compact linear representation capturing dominant modes of terrain variability. Optimal terrain shape coefficients computed using least squares and LASSO regression enabled rapid reconstruction of terrain surfaces. The suitability and effectiveness of these surrogate terrains for operational use were validated through wildfire propagation and plume dispersion simulations using physics solvers FIRETEC and QUIC-PLUME. Results demonstrated that surrogate terrains possess sufficient geometric fidelity for realistic environmental modeling tasks while providing a computationally efficient alternative to full-physics solvers. This surrogate-based strategy is therefore critical for enabling timely, informed wildfire management decisions in operational scenarios.