ECMWF products, used and analysed in this handbook, are encoded in WMO FM-92 GRIB edition 2 excluding Tropical Cyclones. The latter is encoded in WMO FM-94 BUFR format.
Technical details¶
IFS Data¶
IFS open data products are available at 0.25 degrees resolution, one hour after the real-time dissemination schedule. The file names are following this naming convention: ROOT/yyyymmdd/HHz/model/resol/stream/yyyymmdd HH0000-step U-stream-type.format.
Table 1:File-naming convention of the ECMWF IFS model outputs. [1]
| File-naming convention | Values |
|---|---|
| ROOT | URL of a site hosting the open data |
| yyyymmdd | reference date |
| HH | reference time |
| model | ifs |
| resol | 0p25 |
| stream | oper, enfo, waef, wave, scda, scwv |
| step | forecast time step |
| U | unit of the time step (h) |
| type | fc, ef (cf, pf), ep, and tc |
| format | grib2 or bufr |
Both types cf and pf are included in the files with type=ef.
AIFS Single Data¶
AIFS open data products are released as soon as the data are generated. These data are also produced at 0.25 degrees resolution.
Table 2:File-naming convention of the ECMWF AIFS model outputs. [1]
| File-naming convention | Values |
|---|---|
| ROOT | URL of a site hosting the open data |
| yyyymmdd | reference date |
| HH | reference time |
| model | aifs-single |
| resol | 0p25 |
| stream | oper |
| step | forecast time step |
| U | unit of the time step (h) |
| type | fc and tc |
| format | grib2 or bufr |
AIFS Ensemble Data¶
The same as for the AIFS open data products apply to the AIFS ENS.
Table 3:File-naming convention of the ECMWF AIFS ENS model outputs. [1]
| File-naming convention | Values |
|---|---|
| ROOT | URL of a site hosting the open data |
| yyyymmdd | reference date |
| HH | reference time |
| model | aifs-ens |
| resol | 0p25 |
| stream | enfo |
| step | forecast time step |
| U | unit of the time step (h) |
| type | cf, pf, and ep |
| format | grib2 or bufr |
Index files¶
Each GRIB file has its corresponding index file which can be accessed by substituting the .grib2 extension with .index in the URL. Each line in an index file represents a GRIB field in the corresponding GRIB file. It is described using the MARS query language. The keys _offset and _length represent the byte offset and length of the corresponding field respectively. This allows us to download a single field from the GRIB file using HTTP byte-range requests, see Retrieve Data.
Files with the type=ep contain ensemble means and standard deviations as well as probabilities datasets.
The examples below show a line extracted from an index file of the:
- IFS:
{"domain": "g", "date": "20250608", "time": "0000", "expver": "0001", "class": "od", "type": "fc", "stream": "oper", "levtype": "sfc", "step": "0", "param": "tp", "_offset": 14804835, "_length": 224}- AIFS:
{"domain": "g", "date": "20250608", "time": "0000", "expver": "0001", "class": "ai", "type": "fc", "stream": "oper", "step": "0", "levtype": "sfc", "param": "100v", "_offset": 0, "_length": 1382370}- AIFS ENS:
{"domain": "g", "date": "20250803", "time": "0000", "expver": "0001", "class": "ai", "type": "cf", "stream": "enfo", "step": "0", "levtype": "sfc", "param": "2t", "model": "aifs-ens", "_offset": 61637992, "_length": 618016}