*2.2. Meteorological Covariates*

Regional averages of gridded weather variables from the University of Idaho gridMET dataset [47] are used to calculate 12 weather predictors that will provide coarse scale environmental descriptions during each month between 1984–2015. Of these 12 weather predictors, four are measures of temperature; six are measures of moisture levels, and two measure wind characteristics. The four temperature metrics are based on monthly space-time averages of daily average temperature, which are calculated by dividing the sum of the daily maximum and minimum values by two (Weiss et al., 2005 [48]). The quantity hereafter referred to as seasonality measures intra-annual temperature variability by normalizing monthly temperature averages by the mean and standard deviation of all 360 measurements in the most recent 30 years of data (e.g., 1986–2015). The inter-annual temperature variability is captured with a quantity referred to as the departure from normal, which instead normalizes by the mean and standard deviation of 30 measurements in the most recent 30 years of data that correspond to same month as the raw measurement. The remaining temperature metrics are the rolling 12-month minimum and maximum temperature, which will record extreme temperature events that have potential for delayed impacts on wildfire activity. The six moisture level metrics are average specific humidity and precipitation totals over five time periods (1, 3, 6, 12 and 24-month time windows). In addition to a simple space-time average of wind speed at 10 m, the maximum daily space-time average each month was also included as a covariate of the fire occurrence probabilities.
