International Journal of Wildland Fire · 2015 · 45 citations · 64 references
Natural SciencesEarth SciencesMeteorologyClimatologyForecastingWeather ForecastingClimate ModelingFire ModelingFire WeatherFuel DynamicsMedium-range ForecastingSeasonal ForecastingAntecedent Precipitation IndexFire Seasonality AnalysisFire Weather AssessmentFire Weather IndicesFiresSeasonality Of BurningStandardized Precipitation Evapotranspiration IndexStandardized Precipitation Index
In this study we explore the seasonal predictability of summer wildfires in a Mediterranean region (north-eastern Spain), developing a multiple linear regression model with antecedent and current-summer drought indices (Standardised Precipitation Index; and Standardised Precipitation Evapotranspiration Index). This model is based on the assumption that large summer fires in Mediterranean environments are favoured by current-summer drought (proxy for the climatic factors that affect fuel flammability) and by antecedent wet conditions (proxies for the climatic factors influencing fine fuel availability and connectivity). We test three forecast systems based on (i) seasonal European Centre for Medium-Range Weather Forecasts (ECMWF) System-4 forecasts; (ii) persistence and (iii) climatology. These approaches are evaluated through a leave-one-out cross-validation over the period 1983–2012. The climatology forecast, which considers only antecedent wet or dry conditions with a time lag of 2 years, shows some amount of skill in simulating above- or below-normal summer fire activity. ECMWF System-4 proves to be of limited added value with respect to the climatology forecast. Finally, the persistence forecast, which is driven by antecedent conditions and drought conditions just before the start of the fire season, allows more satisfactory results (correlation of 0.49). The results suggest that long-term forecasts of above-normal burned area are feasible in Catalonia (north-eastern Spain), information that could be potentially applied also to other Mediterranean-type regions.
64
R: A Language and Environment for Statistical Computing
R Core Team · 2000 · 353K citations · Full text
Natural Sciences, Computer Science, Statistical Computing +10
Warming and Earlier Spring Increase Western U.S. Forest Wildfire Activity
A. L. Westerling, Hugo G. Hidalgo, Daniel R. Cayan et al. · Science · 2006 · 5.2K citations · Full text