Efforts to delineate the quantitative influence of natural atmospheric variability on regional annual wildfire activity have previously been complicated by the stochastic occurrence of ignition and large fire events, particularly in the fire regimes of Southern and Central California where anthropogenic manipulation is extensive. Traditional methods of quantifying wildfire activity using counts of ignitions and acres burned inherently contain this stochasticity, likely weakening regional fire-climate relationships. In this research, a new method of quantifying regional wildfire activity is developed that aims to more clearly capture the atmospheric fire regime component by aggregating four metrics of fire activity into an annual index value, referred to as the 揂nnual Fire Severity Index� (AFSI), for the 24-year period 1992-2015. Then, using metrics of five weather and climate features known to significantly modulate fire activity in the study region, including the Atlantic Multidecadal Oscillation (AMO), Pacific Decadal Oscillation (PDO), El Nino-Southern Oscillation (ENSO), Santa Ana wind (SAW) events, and coastal marine layer (ML) frequency, the strength of the regional fire-climate relationships contained within the AFSI were evaluated. Singular Spectrum Analysis (SSA), a variant of Principal Component Analysis, was applied to the AFSI time series to identify and attribute the different modes of variability [i.e. Principal Components (揚Cs�)] that comprise it, as resulting from each of the five atmospheric predictors. Analysis of the PC-predictor relationships demonstrated a significant correlation with AMO (r = 0.814) and moderate correlations with PDO, ENSO, SAW, and ML (r = 0.470�630). PDO appears to influence regional fire activity at a longer-term lag of five years, while the influence of ENSO, SAW, and ML frequency is strongest in-year. The weakest PC-predictor relationships were found during periods of concurrent ocean-atmosphere oscillation phase shifts, suggesting that the interrelatedness of AMO, PDO, and ENSO must be better quantified to clarify these fire-climate signals. Overall, our results reinforce the need to rethink how regional fire-climate signals are calculated, and suggest that these methods can be further developed into a predictive model of fire activity.