Wavelet analysis to reveal fluctuations in near-surface temperature and wind speed during the FireFlux I experimental fire

Field-scale wildland fire experiments are needed to validate coupled atmosphere/fire models and to assess their ability to represent wildland fire dynamics, including surface fire spread and perturbed local meteorology in the immediate vicinity of the spreading fire. These experiments provide access to the in situ turbulence observations needed to understand the local turbulent structures, which are impacted by the strong heat fluxes released by the fire and by the presence of forest canopies. In this paper, we present a wavelet analysis approach to extract turbulence information, based on frequency localization, from observed time series of temperature and wind speed during the FireFlux I experimental fire. Wavelets are used to robustly decompose the time series of interest into mean and fluctuation components by separating scales in the frequency space, where it is possible to identify the scales that contribute most to the energy spectrum. Fluctuations in temperature and wind speed are then used as input to a quadrant analysis to characterize the possible interactions between variables as the fire front passes, and to identify the most dominant events such as sweeps, puffing and local cooling. Wavelets therefore have great potential for observation-model comparison, i.e. for assessing the ability of coupled atmosphere-fire models to represent realistic turbulence patterns associated with atmosphere-canopy-fire interactions.