IFS Cycle 50r1 forecast production is currently running in parallel with the existing operational production.
The details of this new model cycle (IFS Cy 50r1) are outlined on the Implementation of IFS Cycle 50r1 page. As it comes with substantial changes to data, configuration requirements and file naming, we strongly encourage all real-time users to test the data.
This page explains the testing process and the actions the users need to take to prepare.
All users, regardless of Service Pack, must test 50r1 data.
It is crucial to test the IFS Cycle 50r1 data, as the cycle upgrade will bring substantial changes to your data, data configuration requirements and file naming.
All Real-Time users need to be prepared.
To get started with testing IFS Cycle 50r1 data, you will need to follow steps outlined below, in the chapter:How To Test
Please be aware that AIFS will be upgraded on the same day as IFS Cycle 50r1. AIFS test data will be available later in the same way. For more details on the changes to AIFS data, please have a look at the following pages: Implementation of AIFS Single v2 andImplementation of AIFS ENS v2.
The Release Candidate Phase for IFS started on 19 Feb 2026. The first cycle available in XDISS was 12Z 18 Feb 2026. File retention duration in XDISS is approximately 5 days. The RCP will allow a test period of at least two months.
To support testing of the new data, a dedicated test environment has been established, which generates IFS Cycle 50r1 test data for all users. Users with Gold and Silver Service Packages are also able to access these test environments.
Submission of the form will create a ticket with the Data Support team, who will make contact to set up or confirm the host configuration for the test data.
Data Support will then send a test file to verify the connection.
Further steps for testing will depend on the specific Service Package held by the user, as outlined below:
Basic + Bronze + Research Users
As Basic/Bronze/Research users do not have access to TPREd or XDISS, all 50r1 test changes to data requirements will be managed by the Data Support team.
A test batch of data will be manually pushed to the users test host for review.
If the user is satisfied, confirmation should be provided via the ticket. This will enable the Data Support team to mark the feeds as “Ready to migrate.”
If operational data requirements are modified prior to the IFS Cycle 50r1 implementation, the process will need to be repeated to ensure that user is prepared for the updated requirements on Implementation day.
Silver + Gold + Member/Cooperating State Users
As a Silver/Gold or MS/CS user, access is available to
If access is not available, please contact the Data Support team via the relevant ticket.
As part of the preparation for testing, ECMWF have already adjusted all data requirements in the test PREd (TPREd) to ensure compatibility with the Cycle 50r1 upgrade for IFS data. Please see the below section 'Key Configuration Changes' for the changes that have been applied.
Once satisfied with the IFS Cycle 50r1 test data, users must mark the feeds in TPREd as “Ready to Migrate” by switching the toggle (located in the top-right corner of the interface) to ON. This action confirms that the TPREd feeds have been tested and are satisfactory to become the operational feeds on the day of implementation. The toggle may be switched off again, and further changes can be made to the feed, until the end of the testing phase.
Users should note that if changes are made to their operational feeds (PREd) during this period, those changes will need to be replicated in the test PREd in order to retain them.
If the “Ready to Migrate” toggle is not enabled, ECMWF will retain the user’s current operational PREd requirements and apply only mandatory system changes on Implementation day.
Please note that changes will affect anyone with the control forecast (cf) for ENS data from IFS. The control forecast (cf) will be in a different file than the perturbed forecast (pf).
This applies only for the control forecast within following streams: stream=enfo and stream=waef.
You can identify the test data by the version number:
IFS Cycle 50r1 test products are available as version 80 (file names include “*80” asexpver).
Operational files keep the file naming convention remains the same as for your operational data.
You will only see the changes you made or requested if you have confirmed that your feeds are "Ready to Migrate". On the day of implementation, the esuite (TPREd) will then become the operational (PREd). This means all the data requirement changes that were done to test your 50r1 data requirements, will start to be delivered operationally. If you have not marked your feeds as 'Ready to migrate', we will update your operational requirements automatically.
If you're a Basic/Bronze user, the Data Support team can push a test batch of IFS Cycle 50r1 data to the host configuration.
If you're a Silver/Gold user, then you can either push the data to yourself or set the data to be delivered in 'NORMAL' mode. If you put the data in 'NORMAL' mode, the data will be disseminated to your host automatically. However, the data are disseminated between operational windows to protect network capacity, meaning the data will not be disseminated according to the usual dissemination schedule. There is no fixed delivery schedule, and products may not be available in real time, delays of up to two days are possible.
If you put your data normal mode, the data will be disseminated between operational windows to protect network capacity. There is no fixed delivery schedule, and products may not be available in real time, delays of up to one or two days are possible. Usually as we get close to implementation the delay gets smaller.
The same TOTP token should be used for the xdiss as for the diss system.
No, the test data are strictly for non-operational use. Test products are intended for internal testing and validation only and must not be used in operational services or shared on social media.
In the test PREd (TPred), you will be able to test as much data as you want (even over your current Volume Band), however when it comes to operational day, if you have made changes that are over your current Volume Band, you will need to do an amendment to your current contract.
Not necessarily. The host configuration of the test system might not be in sync with the operational system. You also might want to receive the data files at a different endpoint. In case, you would like to make changes, please indicate this in the form: RCP 50r1 - Test System User Details or contact us via https://support.ecmwf.int
As a Silver, Gold, or MC/CS user, you can put your feeds in pre-scheduled mode. However, until implementation day, this will function the same as normal mode and you will not receive data pre-schedule. If you would like to start receiving data in pre-scheduled mode after implementation day, we recommend keeping your feeds in this setting. Once you toggle the “Ready to Migrate” button, the pre-scheduled mode will become active then.
IFS Cycle 50r1 beta test datais available from MARS with E-suite experiment version (expver) 0080 (MARS keywords EXPVER=0080, CLASS=OD). If you have access to MARS or if you have an agreement that gives you access to the archive, you can access these data. Archived test data produced before the start of the Release Candidate Phase (before 18Z run), should beused with cautionas it may contain errors that were subsequently fixed. The data should not be used for operational forecasting.
Since the IFS ENS-cf data will be incorporated into the D* files (Set-i), this means the data will be disseminated earlier than the ENS-pf data. For further details, please refer to the Dissemination schedule.
We keep the test files on the test ECPDS for a period of 5 days. Increasing the lifetime of the files on the test ECPDS will not work, as the data are deleted from the HPC after 5 days.
For IFS Cycle 50r1, Ensemble Control forecast (ENS-cf) will become oper-fc and will not have perturbationNumber=0. The plan is to change this in IFS Cycle 50r2, with the GRIB2 migration.
Yes, we will provide re-forecast products to allow creation of a model climate for both medium-range and extended-range. This will be provided in the next few weeks and more information will follow when they are available.
Yes, if you would like to switch to the new file naming convention, you can test it in the test PREd (TPREd) environment during the IFS Cycle 50r1 Release Candidate phase.
If you are already updating your scripts, it would be a good opportunity to switch to the 'ECMWF' file naming convention at the same time.
Yes,AIFS will be upgraded on the same day as IFS Cycle 50r1.
AIFS Single, the deterministic model, will be upgraded from v1.1 to v2. Please refer to Implementation of AIFS Single v2 for details of the implementation of AIFS ENS v2.
AIFS ENS, the ensemble model, will be upgraded from v1 to v2. Please refer toImplementation of AIFS ENS v2 for details of the implementation of AIFS ENS v2.
AIFS test data will be available later under a different experiment version number. Please subscribe toforecast_changes-request@lists.ecmwf.intto receive notifications about updates, including when the test data will be available.
We can already confirm that AIFS enfo-cf data for AIFS will continue to be disseminated.
Also, there will be no dissemination schedule change for AIFS data.
On implementation day, we start using 50r1 as the operational cycle for IFS. You will only see the changes you made or requested if you have confirmed that your feeds are "Ready to Migrate". On the day of implementation, the esuite (TPREd) will then become the operational (PREd). This means all the data requirement changes that were done to test your 50r1 data requirements, will start to be delivered operationally. If you have not marked your feeds as 'Ready to migrate', we will update your operational requirements automatically.
