Datasets

Pangu-Weather: a deep learning-based system developed by Huawei. It is initialised with ECMWF analysis. Pangu-Weather operates at 0.25° resolution.

Pangu-Weather: a deep learning-based system developed by Huawei. It is initialised with ECMWF analysis. Pangu-Weather operates at 0.25° resolution.

Stratosphere relaxation experiment run at Tco199 for 46 days over 20 years starting on 12, 16 , 19, 23, 26 , 30 December and 2nd, 6th and 9th January 1999-2018.

Examples

Retrieving 2-meter temperature of the hindcast starting on 12 December 2019 for all perturbed members.

The climatological bias-correction experiment consists of sub-seasonal hindcasts performed with IFS cycle 48r1, in which a fixed tendency correction is applied to account for systematic model error. Hindcasts are initialised on the 15th of each month from 1989 to 2016 and integrated forward for 32 days. For each start date, an ensemble of 10 members is generated, comprising one unperturbed control and nine perturbed members to sample initial-condition uncertainty. The model is run at TCo319 horizontal resolution with 137 vertical levels.

The control experiment consists of sub-seasonal hindcasts performed with IFS cycle 48r1, without any explicit bias correction or model-error tendencies applied. Hindcasts are initialised on the 15th of each month from 1989 to 2016 and integrated forward for 32 days. For each start date, an ensemble of 10 members is generated, comprising one unperturbed control and nine perturbed members to sample initial-condition uncertainty. The model is run at TCo319 horizontal resolution with 137 vertical levels.

The flow-dependent bias-correction experiment consists of sub-seasonal hindcasts performed with IFS cycle 48r1, in which model-error tendencies are estimated online using neural networks. Hindcasts are initialised on the 15th of each month from 1989 to 2016 and integrated forward for 32 days. For each start date, an ensemble of 10 members is generated, comprising one unperturbed control and nine perturbed members to sample initial-condition uncertainty. The model is run at TCo319 horizontal resolution with 137 vertical levels.

Standalone ecWAM wave model CY50R1 forced by corrected ERA5 hourly neutral 10m winds, air density, gustiness and sea ice fraction (1979-1989). Native grid is Tco1279 (9km), 36 directions, 36 frequencies. See Bidlot et al. 2026: Wave Hindcast for ERA6 Preparation and Training Data Driven Models. ECMWF Tech Memo, in preparation, 6-hourly output, many parameters.

Standalone ecWAM wave model CY50R1 with altimeter wave height assimilation forced by corrected ERA5 hourly neutral 10m winds, air density, gustiness and sea ice fraction (1990-1999). Native grid is Tco1279 (9km), 36 directions, 36 frequencies. See Bidlot et al. 2026: Wave Hindcast for ERA6 Preparation and Training Data Driven Models. ECMWF Tech Memo, in preparation, 6-hourly output, many parameters.

Standalone ecWAM wave model CY50R1 with altimeter wave height assimilation forced by corrected ERA5 hourly neutral 10m winds, air density, gustiness and sea ice fraction (2000-2015). Native grid is Tco1279 (9km), 36 directions, 36 frequencies. See Bidlot et al. 2026: Wave Hindcast for ERA6 Preparation and Training Data Driven Models. ECMWF Tech Memo, in preparation, 6-hourly output, many parameters.

Standalone ecWAM wave model CY50R1 with altimeter wave height assimilation forced by corrected ERA5 hourly neutral 10m winds, air density, gustiness and sea ice fraction (2016-2023). Native grid is Tco1279 (9km), 36 directions, 36 frequencies. See Bidlot et al. 2026: Wave Hindcast for ERA6 Preparation and Training Data Driven Models. ECMWF Tech Memo, in preparation, 6-hourly output, many parameters.

EXPECT seasonal hindcast experiment ConstantCO2/ocean. CO2 emissions kept constant at 1993 levels and warming footprint removed from ocean initial conditions. (For full details see Mayer et al. 2026). Experiment uses Cy49R2 v8 at Tco199 and ORCA025Z75 resolution (the reference ocean ICs are from ORAS6 0001). 51 member ensemble of May initialisations for the years 1993-2026 and run for 7 months.

Examples

EXPECT seasonal hindcast experiment Double delta CO2/ocean. Doubles the forced pattern of climate change in the ocean ICs after 1993 and doubles the annual rate of CO2 increase after 1993 . (For full details see Mayer et al. 2026). Experiment uses Cy49R2 v8 at Tco199 and ORCA025Z75 resolution (the reference ocean ICs are from ORAS6 0001). 51 member ensemble of May initialisations for the years 1993-2026 and run for 7 months.

Examples

11-member coupled subseasonal re-forecast experiment using a prototype CY49R2 configuration of the IFS–NEMO4–SI3 system. Re-forecasts are initialized on the first day of each month, covering the period 2004 to 2023, and are integrated out to +46 days. The atmospheric component runs at Tco319L137 and is initialized from ERA5, ocean and sea-ice components run at eORCA0.25Z75 and are initialized from ORAS6.

Examples

11-member coupled subseasonal re-forecast experiment using a prototype CY49R2 configuration of the IFS–NEMO4–SI3 system. Re-forecasts are initialized on the first day of each month, covering the period 2004 to 2023, and are integrated to +46 days. The atmospheric component runs at Tco319L137 and is initialized from ERA5, ocean and sea-ice components run at eORCA0.25Z75 and are initialized from ORAS6.

Standalone ecWAM wave model CY50R1 with altimeter wave height assimilation forced by operational lwda 6-hourly stress equivalent neutral 10m winds, air density, gustiness and sea ice fraction. Native grid is Tco1279 (9km), 36 directions, 36 frequencies. See Bidlot et al. 2026: Wave Hindcast for ERA6 Preparation and Training Data Driven Models. ECMWF Tech Memo, in preparation, 6-hourly output, many parameters.

Standalone ecWAM wave model CY50R1 with altimeter wave height assimilation forced by operational lwda 6-hourly stress equivalent neutral 10m winds, air density, gustiness and sea ice fraction (2021-2026-02-03). Native grid is Tco1279 (9km), 36 directions, 36 frequencies, See Bidlot et al. 2026: Wave Hindcast for ERA6 Preparation and Training Data Driven Models. ECMWF Tech Memo, in preparation, 6-hourly output, many parameters.

Standalone ecWAM wave model CY50R1 with altimeter wave height assimilation forced by operational lwda 6-hourly stress equivalent neutral 10m winds, air density, gustiness and sea ice fraction AND surface currents from ORAS6 (2016-2020). Native grid is Tco1279 (9km), 36 directions, 36 frequencies, current refraction. See Bidlot et al. 2026: Wave Hindcast for ERA6 Preparation and Training Data Driven Models. ECMWF Tech Memo, in preparation, 6-hourly output, many parameters.

Standalone ecWAM wave model CY50R1 with altimeter wave height assimilation forced by operational lwda 6-hourly stress equivalent neutral 10m winds, air density, gustiness and sea ice fraction AND surface currents from ORAS6 (2021-2026-02-03). Native grid is Tco1279 (9km), 36 directions, 36 frequencies, current refraction. See Bidlot et al. 2026: Wave Hindcast for ERA6 Preparation and Training Data Driven Models. ECMWF Tech Memo, in preparation, 6-hourly output, many parameters.