Downscaled GCM climate projections of fire weather over Victoria, Australia. Part 2*: a multi-model ensemble of 21st century trends.

This paper applies the Multivariate Adaptive Constructed Analogs (MACA) statistical downscaling method to 12 general circulation models to produce 21st century projections of fire weather variables over Victoria, Australia, under two emissions scenarios. The statistically downscaled model data accurately replicate the observed distributions of meteorological variables over the contemporary period, but underestimate fire danger extremes in some models. Under each climate scenario, both mean and extreme fire danger are expected to increase. Though there is variation across Victoria, the 12-model average by year for RCP8.5 indicates a 10–20% increase in extreme (99th percentile) Forest Fire Danger Index across the state, with the greatest change projected in the north-west region. At five geographically and climatologically different locations across Victoria, there is a 50–200% increase in the number of days per year exceeding the threshold for the Victorian Very High or higher fire danger rating by the end of the century compared with the start. The high-end warming (RCP8.5) scenario shows increased temperature to be the main driver of heightened fire danger. Changes in temperature, humidity and precipitation during spring and early summer both increase the length of the fire season and may reduce springtime opportunities for prescribed burning. Twelve GCMs are downscaled over Victoria, Australia, to estimate trends in fire weather frequency under two emissions scenarios using the Multivariate Adaptive Constructed Analogs (MACA) method. Increases of ~50–200% in the number of Very High or above fire danger days are likely by 2100, mostly due to higher temperatures. Photo showing smoke from a planned burn in Victoria, Australia, by Nick McCarthy. [ABSTRACT FROM AUTHOR]