Implementation date: 25.06.2013
Meteorological changes
- Number of vertical levels increased from 91 to 137 in high-resolution forecast model (HRES), the ensemble of data assimilations (EDA), the main assimilation (4DVAR) and the Boundary-Conditions (BC) optional programme.
- Revised background error variances at 137 levels based on IFS cycle 38r1.
- Revised EDA calibration and filtering for 137 levels. Model error cycling in stratosphere switched off.
- Modification of surface drag.
- Modified test parcel entrainment in boundary layer and shallow convection.
- Adjustment of non-orographic gravity wave drag to be consistent with System-4.
- Oxygen absorption correction.
- Revision of Sea-ice/SST quality control over the Caspian Sea.
- Revision of the glacier mask over Iceland.
Meteorological impact
IFS cycle 38r2 increases the vertical resolution of the model throughout the troposphere and stratosphere. It enables a better representation of physical processes, clouds, inversions and vertically propagating gravity waves, for example. The forecast impact in terms of objective verification scores against observations and analyses are summarised in the score card.
Upper-air verification scores
Tropospheric upper-air scores are overall slightly positive in northern hemisphere and mainly neutral for Europe and southern hemisphere. Tropics is mixed with some negative results compared to observations but neutral against analyses.
In the extra-tropics the main negative impacts are for the upper-tropospheric relative humidity (300hPa). The main positive impacts are for geopotential in the lower stratosphere, and to a lesser extent also in the troposphere.
Weather parameters
For precipitation and temperature the overall conclusion is a slight improvement in the extra-tropics and a slight degradation in the tropics. The scores for 10m wind show neutral to slightly positive impact both in extra-tropics and tropics. There is an overall slight reduction of wind speed, most notable in Europe at 12 UTC.
Synoptic impact
No significant differences have been found between the synoptic performance of the pre-operational e-suite and the operational forecast.
Tropical cyclone tracks and intensity have been compared for all Tropical Cyclones available in the research and pre-operational e-suites. There is a slight improvement for the position errors from D+3 onwards, although this is not statistically significant. The impact is neutral for the tropical cyclone intensity.
Ensemble
The ensemble forecasting e-suite has run daily from 10 January onwards (00 UTC forecasts only). In general the e-suite is neutral for the ENS in terms of the spread and ensemble-mean skill for the extra-tropics. In the tropics, spread is also little changed; EM error is improved for 850 hPa temperature, and slightly worse for 850 hPa wind speed.
In terms of probabilistic scores, the CRPS is neutral for 500 hPa height, 850 hPa temperature and wind speed over Europe and the northern extra-tropics. Over the southern extra-tropics e-suite scores are better for 500 hPa height, but worse for days 1-3 for temperature at 850 hPa. In the tropics the e-suite shows a significant improvement throughout the forecast range for 850 hPa temperature, and a small but statistically significant degradation for 850 hPa wind speed, consistent with the changes in the EM error.
Technical aspects
ENS model level definition for L91
New model level definitions are available for L137.
See correspondence between the L91 and L137 model levels is available here.
Changes to the GRIB data
The L137 model level data cannot be converted back to GRIB edition 1 due to limitations in the GRIB vertical coordinate table. See Cycle 37r2 Detail for further information about GRIB edition.
Users should be aware that converting the L137 model level data to GRIB edition 1 with grib_api works without error. However, the resulting GRIB edition 1 header contains incomplete
information about the vertical coordinate parameters (the grib_api key NV and the pv array) and, for grid point fields, the list of the number of points in each latitude (the PL array) is corrupted.
Changes to GRIB headers affecting model level fields only
The increase in the number of vertical levels from 91 to 137 in the high-resolution forecast model is reflected in changes to the GRIB headers, specifically the GRIB 2 Section 4 "Product definition section":
| Section | Octets | grib_api key | Old Value | New value |
|---|---|---|---|---|
| 4 | 6-7 | NV | 184 | 276 |
| 4 | 35-nn | pv | L91 | L137 |
Changes to GRIB headers affecting all fields
The GRIB model identifiers (generating process identification number) for the new cycle will be:
- Atmospheric model ID=143 (previous cycle 142)
- Ocean wave model ID=109 (unchanged)
- Limited-area ocean wave model ID=209 (unchanged)
These 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.
New model output
None.
Release aspects
e-suite experiment number: 0062