Implementation date: 24 June 2010
This new cycle includes a new method for providing initial-time perturbations for the ENS. In this new cycle, differences between members of an ensemble of data assimilations (EDA) will be used instead of the evolved singular vectors to create initial spread between ENS forecast members. Initial-time singular vectors will continue to be used in conjunction with the EDA perturbations.
With the introduction of IFS cycle 36r2, the GRIB API library replaced GRIBEX for encoding GRIB products from the ECMWF deterministic forecasting system, the Ensemble Prediction System (ENS), including the monthly extension to 32 days, and the Boundary Condition Optional Project.
The main changes included in this cycle are:
- Ensemble of data assimilations (EDA), providing initial-time perturbations for the ENS;
- Use of the GRIB API library for encoding of GRIB data;
The new cycle does not include any meteorologically significant changes to the deterministic forecasting system. This new cycle shows clear benefit in terms of the spread of the ENS, particularly in the first few days of the forecast:
- NH 500 hPa geopotential height: small decrease in spread reducing the ENS over-dispersion, slightly improved ensemble-mean and probabilistic objective scores;
- NH 850 hPa temperature: clear increase in spread in the first 48 hours making the ENS spread better tuned, slightly improved ensemble-mean and probabilistic objective scores;
- SH 850 hPa temperature: clear increase in spread in the first 4 days making the system more reliable resulting in better probabilistic scores;
- Tropics 850 hPa temperature: larger increase in ensemble spread significantly reducing the ENS under-dispersion in the Tropics leading to better ensemble-mean and probabilistic scores.
The new GRIB model identifiers are:
- Atmospheric model: 137
- Global wave model: 105
- Limited area wave model: 205