Simulating spatial and temporally related fire weather

Abstract

Use of fire behavior models has assumed an increasingly important role for managers of wildfire incidents to make strategic decisions. For fire risk assessments and danger rating at very large spatial scales, these models depend on fire weather variables or fire danger indices. Here, we describe a method to simulate fire weather at a national scale that captures the same patterns of seasonality, spatial correlation and temporal correlation as the observed inputs. This method incorporates daily surface weather observations, which are incorporated into large scale models that give a grid of modeled weather values across a landscape. We use those observations to obtain estimates of the seasonal trend, the pixel-by-pixel temporal autocorrelation, and the spatial autocorrelation of the remaining residuals. We then use these three sub-models to simulate artificial seasons of weather data. Finally, we verify the results to ensure that the simulated seasons are indistinguishable from the input observations, from a statistical standpoint.

Citation

Grenfell, Isaac C.; Finney, Mark; Jolly, Matt. 2010. Simulating spatial and temporally related fire weather. In: Viegas, D. X., ed. Proceedings of the VI International Conference on Forest Fire Research; 15-18 November 2010; Coimbra, Portugal. Coimbra, Portugal: University of Coimbra. 9 p.