Operational assimilation of Metop-C data

Sean Healy, Niels Bormann, Katie Lean


ECMWF is now assimilating measurements from EUMETSAT’s polar-orbiting Metop‑C satellite, which was launched on 7 November 2018. The new Metop-C data used at ECMWF includes radio occultation (RO) bending angles from the GRAS instrument, radiances from the AMSU-A and MHS instruments, and atmospheric motion vector (AMV) dual-Metop wind information derived from the AVHRR instrument. These changes followed evaluation and testing of the data in the ECMWF system between December 2018 and February 2019.

The Metop‑C satellite joins Metop‑A and Metop‑B in a ‘tristar’ configuration in a morning orbit (i.e. passing over the equator at 09:30 local time). These satellites provide key observations for both global and regional numerical weather prediction (NWP).

Metop-C GRAS data

The addition of Metop-C GRAS measurements significantly increases the total number of RO profiles available for operational assimilation by around 30%, bringing EUMETSAT’s contribution to typically 1,900 RO profiles per day. The EUMETSAT RO team produced high-quality bending angle profiles by 13 November 2018, within six days of launch. The data was then made available to calibration–validation (CAL–VAL) partners, including the EUMETSAT Radio Occultation Meteorology Satellite Application Facility (ROM SAF), for evaluation. It was quickly established that the Metop-C bending angles were of similar quality to both the Metop‑A and B measurements, by comparing them with NWP information mapped to observation space. Assimilation experiments with Metop-C GRAS over a three-month period were then conducted. These showed that the new measurements improved the agreement of ECMWF short-range forecasts with other conventional and satellite measurements, and there was some improvement in stratospheric temperature biases. However, the impact of Metop-C GRAS on medium-range tropospheric forecasts was not found to be statistically significant for this period. Consequently, while assimilation of the data started on 14 March 2019, these assimilation experiments are being extended.

Observation-minus-background departure statistics. The bending angle observation-minus-background (O–B) departure statistics (standard deviation and mean) as a function of impact height for the Metop satellites. The departures are normalised by dividing them by the bending angle noise values used when assimilating the data. The statistics are computed for the period 27 November to 2 December 2018.

Metop-C AMSU-A and MHS data

Metop-C AMSU-A and MHS data were also made available quickly in November 2018, and the data quality was evaluated by comparing observations against equivalent values calculated from short-range forecasts. The AMSU-A comparisons showed that the overall data quality was consistent with that of other AMSU-A instruments currently in orbit. The Metop-C MHS exhibited poorer noise performance when compared to the instruments on the other Metop satellites for two of the three sounding channels, with some variations of the increased noise over time. Whilst this suggested that it might not have the impact of the other MHS instruments, it did not rule out testing the active assimilation of the data. This is because errors in the short-range forecasts (the ‘background’) used in the data assimilation system for humidity are larger, and error contributions other than instrument noise tend to matter more, when we assimilate the data in all-sky conditions.

Short assimilation trials assimilating both AMSU-A and MHS over three months produced a small but statistically significant positive forecast impact. This is particularly remarkable as the instruments were added as the 9th microwave temperature sounder and the 11th microwave humidity sounder in the assimilation. Furthermore, they are the third set of microwave instruments in the 09:30 orbit, already populated by Metop-A and B. Assimilation of the data started on 14 March 2019.

Dual-Metop AMVs with Metop-C

EUMETSAT provides both polar AMVs from a single Metop satellite and global AMVs derived from imagery using pairs of Metop satellites, including Metop-C, known as dual-Metop AMVs. The first polar AMVs derived from AVHRR data on Metop-C became available in early January 2019, and data characteristics for these are comparable to those from Metop-A or B. Assimilation trials are ongoing and activation of the data is expected for later in the year.

The dual‑Metop AMVs – including Metop‑C –  have been used in the operational ECMWF system since 17 January 2019, when the new combination of Metop‑C/B was included alongside the pairing of Metop‑A/B. At ECMWF, the dual-Metop AMV information is used to fill the gap between the traditional geostationary AMVs and the single-satellite polar AMVs, in the latitude bands between 40 and 60 degrees in both hemispheres. A detailed analysis has shown that the dual-Metop AMVs have characteristics that depend on which satellite pair is used. These differences are primarily in the tropics, rather than where the data is used at ECMWF, and they are currently under further investigation by EUMETSAT.

Other instruments

Test data from the Metop-C IASI instrument is expected in April 2019 and will be evaluated in a similar way as the other instruments. Surface wind and soil moisture information retrieved from Metop-C ASCAT measurements is currently being assessed.