Refining Fire–Climate Relationship Methodologies: Southern California

Abstract

Efforts to delineate the influence of atmospheric variability on regional wildfire activity have previously been complicated by the stochastic occurrence of ignition and large fire events, particularly for Southern California, where anthropogenic modulation of the fire regime is extensive. Traditional metrics of wildfire activity inherently contain this stochasticity, likely weakening regional fire–climate relationships. To resolve this complication, we first develop a new method of quantifying regional wildfire activity that aims to more clearly capture the atmospheric fire regime component by aggregating four metrics of fire activity into an annual index value, the Annual Fire Severity Index (AFSI), for the 27-year period of 1992–2018. We then decompose the AFSI into trend and oscillatory components using singular spectrum analysis (SSA) and relate each component to a set of five climate predictors known to modulate macroscale fire activity in Southern California. These include the Atlantic Multidecadal Oscillation (AMO), Pacific Decadal Oscillation (PDO), El Niño–Southern Oscillation (ENSO), and Santa Ana wind (SAW) events, and marine layer frequency. The results indicate that SSA effectively isolates the individual influence of each predictor on AFSI quantified by generally moderate fire–climate correlations, |r|>0.4, over the full study period, and |r|>0.5 over select 13–15-year periods. A transition between weaker and stronger fire–climate relationships for each of the oscillatory PC–predictor pairs is centered around the mid-2000s, suggesting a significant shift in fire–climate variability at this time. Our approach of aggregating and decomposing a fire activity index yields a straightforward methodology to identify the individual influence of climatic predictors on macroscale fire activity even in fire regimes heavily modified by anthropogenic influence.