Did the models place the large-scale trough and ridge axes in the correct locations?
The models showed useful skill with respect to this problem February 1 to February 5 including during the largest event as a negative-tilt trough came onshore February 3 (Fig. 6). Both models did poorly the last three days of the evaluation on February 6-8. An ongoing NCEP/WR evaluation of these models has shown: (1) this variability in skill is typical at this forecast range; and (2) the MRF has, for winter 97/98, been better at forecasting changes in the large-scale pattern at the 120-hour range. Figure 7 indicates the MRF and UK models scored better than NOGAPS but worse than the ECMWF model over the period December 1, 1997 through February 28, 1998. Q2. Did the models forecast the amplitude of the troughs as they reached California? No. At 120 hours both models tended to under-forecast amplitude. 2. MRF ensemble 132-hour 500mb height spaghetti charts Note: The MRF ensemble consists of 17 forecasts. Each forecast begins with slightly different initial conditions. The spread of the 1
