Creation of SUBS-M-Climate

The SUBS-M-Climate is derived from a set of sub-seasonal range re-forecasts created using the same calendar start dates over several years for data times either side of the time of the sub-seasonal ensemble run itself.  The re-forecast runs are at the same resolution as the sub-seasonal medium range run (currently 36km) and run over the 46-day sub-seasonal range ensemble period. 

There is merit in examining the real-time performance of a forecasting system.   But the sample sizes created for one system are far too small to conclude anything about its true performance levels.  Re-forecasts are used to increase the available data to produce a model climate.   The results of forecast system may be compared with this model climate.

Re-forecasts are a fundamental component of sub-seasonal forecasting system; they have two applications:

  • sub-seasonal range forecast verification metrics are based on the re-forecasts.
  • re-forecasts allow computation of the SUBS-M-climate which allows actual forecasts to be converted into an anomaly format.   Forecasts in terms of anomalies relative to a model climate (rather than relative to the observed climatology) mean that some calibration for model bias and drift into the products is incorporated.

Selection of sub-seasonal range re-forecasts

Re-forecasts are made every two days during each month on 1/3/5/7/9/11/13/15/17/19/21/23/25/27/29/31 (excluding 29 February).  These are the re-forecast base-dates for sub-seasonal range.

The set of re-forecasts for the SUBS-M-climate is made up from:

  • five consecutive re-forecasts centred on the M-climate base date and covering a forecast period of 9 days.  The middle re-forecast corresponds to the closest base-date on or preceding the actual ensemble run date (e.g. if the actual ensemble run date is the 20th of the month, then the re-forecast base-date is the 19th and the re-forecasts used are for 15th, 17th, 19th, 21st, 23rd). 
  • re-forecasts are derived by re-running an 11-member ensemble (1 control and 10 perturbed members) using the same calendar start date for each of the last 20 years.

Thus there are 20 years x 5 re-forecasts x 11 ensemble members = 1100 re-forecast values.  These are available at forecast intervals of 6 hours for each forecast parameter, for each forecast lead-time and for each calendar start date and location.  They are used to define the SUBS-M-climate.  

Running on fixed days of the month allows direct comparison between re-forecasts produced at different resolutions and/or in different years.


The SUBS-M-climate is used in association with the sub-seasonal range ensemble forecast:

  • to present the day15 to day46 ensemble meteograms with the sub-seasonal range climate (SUBS-M-climate).
  • to highlight significant anomalies of forecast 2m temperature, wind speed, cloudiness and precipitation from the norm for a given location and time of year.  

Running on fixed days of the month allows direct comparison between re-forecasts produced at different resolutions and/or in different years.

Different priorities for M-Climate and SUBS-M-Climate

ECMWF uses different numbers of re-forecasts to build the M-Climate and the SUBS-M-Climate; a smaller number is used for the SUBS-M-Climate, which, because the subseasonal reforecasts are also more closely spaced, means that the "date range" spanned by the M-Climate is about 3-4 times larger than the date range spanned by the SUBS-M-Climate. This is because priorities, with regard to model climate usage, differ for different classes of lead time:  

  • For the medium range, the priority is the best possible capture of the climatological distribution of the tails (e.g. for the extreme forecast index (EFI) and the shift of tails (SOT)).  
  • For sub-seasonal ranges, the priority is better representation of seasonal cycles (for example in the Indian monsoon onset time).


(FUG associated with Cy50r1)