Spatial/temporal analysis of fire spread modeling errors

Advanced remote sensing systems now monitor fires with greater accuracy and at higher sampling frequencies, thus enabling closer analysis of the spatial and temporal characteristics of fire spread modeling errors. A side effect of such systems is the introduction of high frequency variability that is characteristic of noisy data. We describe a method of smoothing such data to focus the analysis at selected resolutions. The method is applied to the analysis of spread simulations of a fire in San Diego County in 2002. Weather data for the simulations were generated from the MM5 mesoscale model at various grid spacings. The effects of variable grid spacing and data smoothing on fire spread modeling effectiveness are discussed.