Description of the upgrade

The vertical resolution is increased from 62 to 91 levels in ENS medium-range and extended-range forecasts.

Atmosphere-ocean coupling now active from initial forecast time in ENS. Various changes made to the parametrisation of physical processes, and to observational data and data assimilation.

One of the key impacts is a dramatic improvement in the diurnal cycle of convection so that peak precipitation occurs later in the afternoon/early evening than in previous cycles.

Implemented: 19 November 2013  

Datasets affected

  • HRES
  • ENS


Resolutions in bold increased/changed from previous IFS cycle.



  • HRES: T1279
  • ENS: Leg A T639, Leg B T319, Leg C T255

  • HRES: 0.25 degrees
  • ENS: 0.5 degrees


  • HRES: L137
  • ENS: L91 (prev L62)

Meteorological content

Vertical resolution of ENS

The vertical resolution and the vertical extent used for the medium-range and monthly ensemble forecasts will change: the number of levels of the ENS will increase from 62 to 91 with the model top raised from 5 hPa to 0.01 hPa. The pressure levels remain unchanged.

Atmosphere-ocean coupling now active from initial forecast time for ENS

The atmosphere-ocean coupling of the ENS will be active from initial time of the forecast using a new version of the NEMO ocean model.

These changes do not apply to the Long-range (SEAS) forecast.

Initial condition perturbations for ENS and EDA

Introduction of perturbation of land surface initial conditions in ENS and perturbation of land surface temperature and moisture observations in EDA.

Changes to physical processes

  • Modifications to convection to address the diurnal cycle of precipitation (ECMWF Newsletter No. 136, pages 15–22).
  • A package of changes introduced to vertical diffusion in stable conditions, turbulent orographic drag, orographic gravity wave drag and surface-atmosphere coupling over forests, which improves boundary layer winds (e.g. at wind turbine hub height) and improves northern hemisphere winter scores (ECMWF Newsletter No. 138 , pages 24-29).
  • An error in the handling of snow albedo in the radiation scheme is corrected.

Data assimilation

  • Dynamic estimation of background error covariances for 4DVAR with enhanced 25-members EDA
  • SSMIS 183 GHz channels activated in all-sky microwave radiance assimilation
  • Enhanced use of AMSU-A, AMSU-B and MHS data over sea ice
  • Situation-dependent observation errors and revised quality-control for AMVs
  • Use of calibrated EDA spread in radiance space as background error estimate for the FG-check for ATOVS


  • Changes to shallow water calculation and surface air density

Meteorological impact

Upper air

The new cycle significantly improves the performance of HRES in the northern hemisphere, especially during autumn/winter time, and it has a neutral to slightly negative impact in the southern hemisphere. The temperature and humidity forecasts are also significantly improved in the lower troposphere in the tropics, while the 850 hPa winds are slightly degraded in certain tropical regions. This issue will be addressed in a forthcoming cycle. ENS and model changes associated with Cycle 40r1 produce overall improvements in probabilistic scores, except for a slight deterioration of tropical and southern hemisphere winds. The inclusion of the EDA-based land-surface temperature and moisture perturbations in ENS improves reliability, especially in the short range.

Weather parameters

The diurnal cycle of convection over land has been dramatically improved (see ECMWF Newsletter 136 - pages 15-22) so that the peak precipitation occurs later in the afternoon/early evening than in previous cycles. Verification of wind speed at a few tall tower locations in Europe has shown that the night time winds have improved in the height range from 50 to 200m, which is relevant for wind energy applications. There is a slight deterioration of ENS temperature in the extratropics. The overall impact on total cloud cover is neutral in the tropics and slightly positive in the extratropics. Precipitation is neutral in the HRES and significantly improved in the ENS for the tropics. Ocean coupling from day 0 in the ENS leads to better SST prediction in cases of slow tropical cyclone propagation.

Monthly forecast

Cycle 40r1 has a statistically significant positive impact on the monthly forecast skill scores in the stratosphere due to the increased vertical resolution and on the prediction of the Madden Julian Oscillation thanks to the ocean-coupling from day 0. The impact on the other monthly skill scores is generally neutral.

Synoptic aspects

The diurnal cycle of convection is much improved and this is apparent in the associated forecast fields including convective indices (CAPE, CIN), precipitation and simulated satellite imagery. The 24-hourly precipitation totals are not significantly affected by the change in timing of the convection. The snow analysis has been improved, while perturbations to snow cover in the ENS have a noticeable effect on 2m temperature spread. Users will also notice the effect of the ocean coupling from day 0 in the ENS on the evolution of the SST.


Score card

New and changed parameters

New parameters


Changes to existing parameters


Technical content

ENS model level definitions

See ENS model level definitions for L91 and the correspondence between the L62 and L91 Most of the additional levels are added in the stratosphere. Levels below ~153 hPa are the same in the two representations (i.e., the 47 levels from 16 to 62 in the L62 representation map identically to levels 45 to 91 in the L91 representation).

See the  list of all other model level definitions currently in use

Changes to the GRIB data

ENS model level fields only

The increase in the number of vertical levels from 62 to 91 in the ensemble forecast model is reflected in changes to the GRIB headers, specifically the GRIB 2 Section 4 "Product definition section":



grib_api key

Old value

New value


Changes to GRIB headers affecting all fields

The GRIB model identifiers (generating process identification number) have changed as follow:


Old ID

New ID

Ocean wave109110
Limited-area ocean wave209210

The GRIB model identifiers are found in:

GRIB 1: Product Definition Section 1, Octet 6
GRIB 2: Product Definition Section 4, Octet 14

or with the grib_api key generatingProcessIdentifier.

Model output

Discontinuation of the following

  • DCDA Atmospheric model (delayed cut-off)
  • DCWV Wave model (delayed cut-off)
  • ENDA Ensemble data assimilation - atmospheric model
  • EWDA Ensemble data assimilation - wave model(from 20/08/2013)
  • Ensemble forecast tube products

Release aspects

e-suite experiment number: 0063 (from 20/08/2013)


ECMWF Newsletter: See Newsletter 138