Abstract
The early 1980s ushered in rapid advances in forecasting skill with the introduction of the Met Office 15-level model in 1982, including much improved forecasting of cyclogenesis. Nevertheless, there remained many opportunities for forecasters to add value to computer output, for example by using an increasing range of satellite imagery. However, failure to predict the impacts from the Great Storm of October 1987 forced a major rethink of the severe weather warnings strategy. Outputs from other forecasting centres became available, in particular, ensembles from European Centre for Medium-Range Weather Forecasts. New operational working procedures evolved to maximise benefits derived from this ever-increasing volume of information. Figure S1. (a) Infra-red image and global model 6-h forecasts of sea-level pressure valid 1200 utc 27 November 2002 over the eastern Pacific Ocean. Bold red arrow shows an observation of 980hPa which was 12hPa below model expectations. (b) Model forecast and observations for same time on the Met Office ‘intervention’ display screen. Observations differing from the model by >3hPa are shown in red (also circled in red). The wind plotted in cyan has been corrected from an assumed erroneous value. (c) As (b) but with cyan plots including additional ‘bogus’ observations created to attempt to replicate the correct position of the low, as inferred from imagery and observations shown in (b). (d) Imagery and observations for 1200 utc 27 November 2002 with original model 6-h forecast (red) and new analysis following intervention (cyan). (e) As (d) but for 1800 utc 27 November 2002 showing that the benefit of intervention has been retained in the new forecast, with the new forecast (cyan) placing the more vigorous depression centre correctly in the dry slot. Crown copyright. Figure S2. (a) Individual solutions (postage stamps) from the ECMWF ensemble for 1200 utc 17 December 1997 from a forecast initiated at 1200 utc 10 December 1997. (b) Plume of 850hPa temperatures at Herstmonceux (Sussex) up to 10 days ahead from the same ensemble forecast showing the spread of solutions, which typically increases with time. The bold dashed line represents the higher resolution deterministic model solution and the bold solid line the unperturbed (control) run at the lower resolution of the ensemble members. Contours represent probability for 1.0 degC intervals, with increasing density of stippling for successive intervals; 0.5–10%, 10–30%, 30–50% and 50–100%. The deterministic run is an outlier, falling outside the range of the ensemble around days 7 to 8. (c) is the ‘central cluster’, effectively a mean of the most common solutions in the ensemble at 1200 utc 17 December 1997, (d) and (e) are ensemble members representative of minority solutions verifying at the same time. Data S1 Supporting Information Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have