Specification of monthly mean anomalies of fire weather elements in the United States

Abstract specification of monthly mean surface weather elements from concurrent fields of 700 mbar monthly mean height anomaly can help the US Forest Service to evaluate the potential for wildland fires. Multiple regression equations, therefore, were derived for anomalies of monthly mean precipitation frequency in 60 climate divisions and of midday temperature, dewpoint, and wind speed at 116 surface stations in the contiguous US. This was done by applying a forward selection (screening) procedure to the simultaneous 700 mbar height anomalies at 79 grid points, for precipitation frequency, and at 121 grid points, plus the previous month's local persistence, for the other weather elements. The cut-off criterion for selecting equations was determined by Monte Carlo simulations plus several additional considerations. Averaged over all months and stations, these specification equations explained 68% of the temperature variance by means of four variables. The corresponding statistics for dewpoint were 57% and three terms, for precipitation frequency 41% and three terms, and for wind speed 37% and two terms. As large geographical and temporal variations were evident in these statistics, maps of the explained variance were analyzed for each element and each month. Graphs of the annual cycle were also prepared for each element for the explained variance, the 1-month lag autocorrelation and the standard error of estimate. These maps and graphs should help the Forest Service to interpret the final specification equations, apply them to prognostic height anomalies prepared by the National Weather Service and issue monthly fire weather outlooks.