Point process modeling of wildfire hazard in Los Angeles County, California

Abstract

The Burning Index (BI) produced daily by the United States government's National Fire Danger Rating System is commonly used in forecasting the hazard of wildfire activity in the United States. However, recent evaluations have shown the BI to be less effective at predicting wildfires in Los Angeles County, compared to simple point process models incorporating similar meteorological information. Here, we explore the forecasting power of a suite of more complex point process models that use seasonal wildfire trends, daily and lagged weather variables, and historical spatial burn patterns as covariates, and that interpolate the records from different weather stations. Results are compared with models using only the BI. The performance of each model is compared by Akaike Information Criterion (AIC), as well as by the power in predicting wildfires in the historical data set and residual analysis. We find that multiplicative models that directly use weather variables offer substantial improvement in fit compared to models using only the BI, and, in particular, models where a distinct spatial bandwidth parameter is estimated for each weather station appear to offer substantially improved fit.

Spatial and temporal pattern of wildfires in California from 2000 to 2019

“…Within the temporal analysis, the wildfire frequency and burned area were plotted with year. Then the segmented linear regression was implemented on these plots to show the trend of wildfires 37 . The coefficient of determination ( R 2 ) and the p-value (p) were added to the plots to indicate the goodness of fit of the regression equation.…”

County‐level analysis of the impact of temperature and population increases on California wildfire data

“…Of course, wildfire ecology is extremely complex (Pyne et al ., ; Johnson and Miyanishi, ; Moritz et al ., ), and several important covariates, such as windspeed, precipitation, drought indices, bark beetle populations, cloud cover, and so on, may act as confounding factors in our analysis. Future work may include the use of Burning Index (BI) and/or the variables determining the BI as covariates, although some previous analyses have shown the BI not to be an accurate predictor of wildfire incidence or area burned in Southern California (Schoenberg et al ., ; Schoenberg et al ., ; Xu and Schoenberg, ). In addition, the model with total area burned as the response had highly skewed residuals and several highly influential observations.…”

The application of prototype point processes for the summary and description of California wildfires

“…Further, for the problems of wildfire forecasting and the estimation of wildfire burn probabilities based on meteorological variables, which are important for urban planning and for the preparation of wildfire suppression and emergency response, a key component is the precise delineation of wildfire seasons, since wildfire hazard models often involve different parameters for different seasons (see e.g. Xu and Schoenberg, 2011). Prototypes may thus be quite directly useful for wildfire forecasting.…”