Ensemble mean and spread: four standard parameters
On this page you can visualise output from the ECMWF 'Ensemble Prediction System' (ENS), for four parameters: mean sea level pressure, 850 hPa temperature, 850 hPa wind speed and 500 hPa geopotential height.
These charts are updated once every 12 hours at approximately 08:30 UTC and 20:30 UTC. Each chart header is labelled with the date and time when the ensemble forecasts were initiated (D0), which will be 00UTC for the 08:30 UTC update, and 12UTC for the 20:30 UTC update. Each map is then valid for a date between D0 + 1 and D0 + 10days, which is indicated in the chart header by VT (=Valid Time) and which can be adjusted using drop down menus above the plot (grey boxes). Additional drop down menus here enable the user to change parameter and geographical region.
Ensemble forecasts explained
One 'ensemble forecast' consists of 51 separate forecasts made by the same computer model, all activated from the same starting time. The starting conditions for each member of the ensemble are slightly different, and physical parameter values used also differ slightly. The differences between these ensemble members tend to grow as the forecasts progress, that is as the forecast lead time increases.
The four parameters that can be selected for display here are listed below. Contouring in blue and red shows absolute values of these parameters (the meaning of the coloured shading is explained further down):
- Pressure reduced to mean sea level (labelled in hPa, 5hPa interval)
- Wind speed on the 850-hPa isobaric surface, which is around 1500m above sea level (labelled in m/s, 5m/s interval).
- Temperature on the 850hPa isobaric surface, which is around 1500m above sea level (labelled in C, 4C interval).
- Height of the 500-hPa isobaric surface, which is around 5.5km above sea level (labelled in dm, 6dm interval).
Note that for 850-hPa wind plots the data resolution used is lower than it is for the other variables - this is to improve legibility.
On the left panels the contours represent the average (mean) of one ensemble - known as the 'ensemble mean'. On the right panels the contours show, for comparison, the single higher resolution forecast.
The spread within the ensemble is also represented on each panel, using coloured shading. On the right panels spread is simply represented as the standard deviation. The units are the same as in the parameter list above. The left panel, meanwhile, shows the 'normalised standard deviation'; to compute this we take the standard deviation from the right panel and divide it by a 'mean standard deviation'. This 'mean standard deviation' is a pre-computed field, that is a function of lead time and of geographical location. If the initiation time of the selected forecast is 00 UTC it represents the mean of the standard deviations of the 30 most recent 00 UTC ECMWF ensemble forecasts. Similarly for a selected 12UTC forecast the 30 most recent 12 UTC forecasts are used.
So, as an example, if the spread at day 5 in a particular ensemble forecast set (right panel) seems to be large, but has of late also tended to be equally large at day 5 in the same area, then the left panel shading will denote a value that is close to 1 (i.e. no colouring). Conversely if the spread in a particular area at day 5, in one ensemble, exceeds the spread that had recently been seen there at day 5, then the left panel shading will indicate a value rather greater than 1 (purple shading).
Relatively large absolute values of standard deviation (right panel shading) tend to indicate relatively high uncertainty in forecasts of the displayed parameter, and thus indicate the extent to which absolute values of the contoured forecast fields (either the ensemble mean on the left panel, or the high resolution forecast on the right panel) can be relied upon. In other words small values on the shaded fields (no colour, or the paler purples) mean high confidence, large values (brighter purples/magentas) mean low confidence.
The normalised standard deviation (left panel shading) aims to put the standard deviation measure into the context of the general ensemble behaviour, in that area, over the last 30 days. So although the forecast for, say, mean sea level pressure on day 8 will ordinarily be a low confidence forecast, there will nonetheless be some occasions when one can be a bit more confident than 'usual', and the normalised standard deviation will tend to show this (by having a value rather less than 1, i.e. green shading).