...
In the following table, it is shown the information about the ensemble sizes, start dates and production schedule for the seasonal forecasting systems contributing to C3S.
| SYSTEM | FORECASTS | HINDCASTS | |||
|---|---|---|---|---|---|
| ENSEMBLE SIZE and START DATES | PRODUCTION | ENSEMBLE SIZE and START DATES | PRODUCTION | ||
| ECMWF | System 4 | 51 members start on the 1st | real-time | 15 members start on the 1st | fixed dataset |
| SEAS5 | 51 members start on the 1st | real-time | 25 members start on the 1st | fixed dataset | |
| Météo-France | System 5 (a) | 51 members 26 start around the 22th | real-time | 15 members start dates? | fixed dataset |
| System 6 | 51 members 1 starts on the 1st | real-time | 25 members 1 starts on the 1st | fixed dataset | |
| MetOffice | GloSea5 (b) | 2 members every day (c) | real-time | 7 members on the 1st | on-the-fly produced around 4-6 weeks in advanced |
| CMCC | SPSv3 (d) | 50 members start on the 1st | real-time | 40 members start on the 1st | fixed dataset |
| DWD | GCFS1 | 50 members start on the 1st | real-time | 30 members start on the 1st | fixed dataset |
(a) Despite they are produced in a lagged mode, the data from MeteoFrance system5 is produced and archived as if all the members were initialized on the 1st
...
(d) CMCC seasonal forecast system routinely produces 80 real-time forecast members but just 50 of them are made publicly available as their contribution to C3S
Description of the c3s-seasonal dataset in the MARS archive
Data tree
In the following sections, a brief description of the contents will be presented. Usually in addition to the explanations about what can be found in every menu, you will find some notes about the MARS keywords involved, i.e., about those elements that are required to define the code that it will be used to retrieve the requested data.
Currently, just the original and post-processed data from all the individual contributors is made available using this data service. Products based on the multi-system combination are just currently available as graphical products
"Multi-model seasonal forecast" (STREAM=MMSF)
The original data in daily and subdaily frequencies for all the individual members from each one of the forecasting systems. Both forecast and hindcasts are included here
"Multi-model seasonal forecast atmospheric monthly means" (STREAM=MSMM)
Monthly aggregated values (average, minimum, maximum and standard deviation) for the individual members. Both forecast and hindcas are included here.
Additionaly it is included here the ensemble mean of the real-time forecasts and the mean of the hindcasts over the complete climate period.
"Multi-model seasonal forecast monthly anomalies" (STREAM=MMSA)
Monthly anomalies with respect to every model's climate (hindcast mean) for the individual members and the ensemble mean
Record of system changes in the contributions to C3S
In the following table, the values of the MARS keyword SYSTEM for the operational C3S seasonal contributors are shown. The systems updated/changed at every row are highlighted in blue colour.
...
from 20171101
onwards
...
(*) Due to some peculiarities of GRIB formatting and MetOffice forecasting system design, the value of the MARS keyword SYSTEM is changed on a yearly basis, so it has no direct relation with the model version of their system
High frequency data (stream MMSF)
TYPE=fc (hidden)
YEAR (note, hindcasts formally equivalent to forecasts)
MONTH
ORIGIN (ecmf, lfpw, egrr)
SYSTEM (hidden, usually... always [are we going to have at some point different systems for the same start date... e.g. MetOffice]?)
LEVTYPE (sfc, plev)
DATE (it is hidden for ecmf, lfpw and 1st of the current month for egrr)
Leaf contents: number, step (freq= 6h, 24h, 12h), [level], parameter
Monthly means and other monthly statistics (stream MSMM)
Derived probability products
TYPE (em, ensemble mean; hcmean: hindcast climate mean, one for every real-time forecast)
ORIGIN
SYSTEM (optional)
LEVTYPE (sfc, plev)
Leaf contents: date, time***, fcmonth, [level], parameter
*** Ask Manuel, why time is not "hidden" in the final leaf (as it is for MMSF)
Derived forecast products
TYPE (fcmean/max/min/stdev: "Forecast mean,...")
YEAR (note, hindcasts formally equivalent to forecasts)
ORIGIN
SYSTEM (optional)
LEVTYPE (sfc, plev)
Leaf contents: date, time***, fcmonth, number, [level], parameter
*** Ask Manuel, why time is not "hidden" in the final leaf (as it is for MMSF)
Real-time forecasts monthly anomalies (stream MMSA)
*** Note: check EUROSIP documentation (it is said that MMSA is not calculated for "lagged" ensembles)
Derived probability products
TYPE (em, ensemble mean)
ORIGIN
SYSTEM (optional)
LEVTYPE (sfc, plev)
Leaf contents: date, time***, fcmonth, [level], parameter
*** Ask Manuel, why time is not "hidden" in the final leaf (as it is for MMSF)
Derived forecast products
TYPE (fcmean "Forecast mean")
ORIGIN
SYSTEM (optional)
LEVTYPE (sfc, plev)
Leaf contents: date, time***, fcmonth, number, [level], parameter
*** Ask Manuel, why time is not "hidden" in the final leaf (as it is for MMSF)