Laserfiche WebLink
<br />D. storm Conditions <br /> <br />It was believed that the forecasting task would differ <br />depending en the weather conditions, and Table 4 presents <br />the results for the breakout of "Non storm Day" versus <br />"storm Day" based on Heggli and Rauber's (1988) storm <br />classification. We quickly see that the CPOD's are <br />uniformly higher for storm days than for the non storm <br />days. In short, when there was a storm on the barrier the <br />forecasters were able to better predict the concentration, <br />onset, and duration of SLW than when there was no sto~. <br /> <br />However, the picture is less clear for the association <br />measure (TSS). Although, there continues to be evidence of <br />a sharp drop in association after the 2nd two hour forecast <br />period in concentration for both categories there is little <br />support for better forecasting of concentration of SLW on <br />"storm" compared to "Non-Storm Days." <br /> <br />~he comparison of onset versus duration predictions <br />follows the previous results. Onset predictions continue to <br />exhibit more skill than duration irrespective of storm <br />category. For "storm Day" the overall skill in predicting <br />onset is almost twice that for duration (e.g., TSS = .67 <br />versus .36). <br /> <br />E. Frontal Type <br /> <br />The question of how well forecasters were able to <br />predict the frontal type as it approached the ARB is <br />addressed in Table 5. Five frontal categories were used <br />(i.e., none, cold, split, cutoff, and other), and all <br />forecasts were re-exar.ined and a few re-coded to eliminate <br />multiple types and match the Heggli and Rauber storm <br /> <br />11 <br />