...
Same as above but for e-suite:
The band of the strongest wind swept over the oil fields on the North Sea. Therefore we have a relatively high density of wave height observations for this case, as seen in the first plot below.
The two plots below compares the o-suite (left) and e-suite (right) significant wave height against observations. The forecast is for 29 January 12z and verified 6 hours later. In the area of highest waves we have 22 observations.
The next plot shows time-series of observations (black), o-suite (red) and e-suite (blue) for significant wave height (left) and 10-metre wind speed (right) for the three last forecasts before the event. The observations are from a 0.5x0.5 degree box centres on 61.2N, 1.1E (3 stations within the box). Note that the wind observations could be higher than 10 metres. Both the maximum wave height and wind speed is underestimated in the forecasts. Note that the wave height is somewhat higher in all three e-suite forecasts compared to o-suite while the wind speeds are similar.
The next plot shows the time-series for the most extreme observation of wind speed (Krakenas). The model clearly missed the peak.
3.3 ENS
The series of plot below shows cyclonic features where the colour indicates the maximum wind at 300 km radius.
...
To see it more quantitatively the evolution of the forecasts, the values for a box around the strongest windgusts observed as [64.5, -5, 56.5, 9.5] including most of Scotland, the Shetland Islands and SW of Norway. The following values are area averages over this box from the HRES, the ENS-control and the Q0, Q25, Q59, Q75 and Q100 of the ENS. The observation is unknown, but the shortest range forecast's 34-35 m/s looks realistic based on the observed wind gust values. The HRES even up to the 00z run on the 24th looks pretty high, only slightly less windy than the latest run on the 29th. In earlier runs the values started to decrease bit faster. All in all the model was able to get continuously closer to the very high values without any major jumpiness, at least on this larger areal average scale. In terms of difference between the operational and e-suite they look very similar. The only maybe systematic difference I can see is the median of the ENS where the e-suite looks higher than the median of the oper forecasts almost in all runs making it slightly better.
The plots below show the ensemble forecast of waves for 61.2N, 1.1E for the forecast initialised 27 January 00z. O-suite (left) and E-suite (right). The waves area clearly higher in the e-suite during the peak period.
3.4 Monthly forecasts
3.5 Comparison with other centres
...