Fire‐Atmosphere Coupling Starts Upwind: Selective Redistribution of Turbulence at the Fire‐Induced Rear Inflow

Abstract Wildfires alter local atmospheric conditions, yet the spatial extent and frequency‐dependent characteristics of fire‐induced inflow perturbations remain largely unquantified. This study applies Multivariate Empirical Mode Decomposition (MEMD) to analyze high‐frequency horizontal wind speed data from a tower positioned at varying distances (∼80–410 m) upwind of three wind‐driven experimental burns in gorse shrub fuels in Rakaia Valley, New Zealand. Measurements at 3‐m, 15‐m, and 30‐m above ground level (AGL) captured rear‐inflow variability across three distinct fire‐atmosphere regimes characterized by ambient wind speeds and flame heights. Wind speed was decomposed into intrinsic mode functions (IMFs) to characterize the multiscale energy distribution of turbulence pre‐, during, and post‐fire, across temporal scales from seconds to hours. This work introduces the novel finding that fire‐atmosphere coupling manifests upwind as selective spectral redistribution rather than uniform turbulence amplification, revealing how fires reorganize atmospheric energy across specific temporal scales. Fire‐induced perturbations decayed with distance but remained detectable hundreds of meters upwind, with the strongest amplification in high‐frequency bands (3–10 s time periods) at 3‐m AGL. During fire, MEMD revealed enhanced spectral energy in shorter bands (∼15 s–2 min), while longer inflow‐scale patterns (∼2.5–30 min) emerged from extended pre‐ and post‐fire analysis. Vertical velocity statistics independently confirm that buoyancy‐driven turbulence is a first‐order control on this spectral redistribution. These findings demonstrate that fire‐atmosphere feedbacks are fundamentally scale‐selective, with important implications for improving parameterizations of fire‐driven turbulence in coupled models.