...
3. Predictability
3.1 Data assimilation
...
Until the last moment the handling of the storm by the forecast system was a challenge. The HRES model initiated on 13@18 shows initial position errors larger than 100 km when compared with the estimated position by the National Hurricane Centre (the average position error is slightly below 50 km). During the next 3 hours the forecast track deviates from the observed one by 150 km (see slide below). A quick look at the ascat winds available around this time shows there was no direct hit on the storm in the Atlantic which could benefit the re-positioning of the storm in the analysis.
MSLP forecast initiated on 13@18 valid at t+0h (left) and t+03h (right). Blue Dots indicate Leslie position at 13@18 and 13@21 estimated by NHC
3.2 HRES
The plot below show observed 24-hour wind gusts (first plot, not sure all obs are in here!) valid for 13 October 12UTC to 14 October 12UTC and HRES forecasts for 24-hour maximum wind gusts (shade) and MSLP valid 14 October 00UTC (contour). Only the HRES from 12 October 12UTC and later predicted well the landfall on Portugal.
...