The European Space Agency’s Aeolus mission has been demonstrated to be a success as the first space-based Doppler Wind Lidar mission, by providing wind observations of good enough quality to improve weather forecasting. This conclusion was reached by comparing the Level-2B (L2B) horizontal line-of-sight (HLOS) wind observations to the ECMWF NWP model equivalents and by the positive impact of Aeolus in Observing System Experiments (OSEs).
The L2B Rayleigh-clear HLOS wind (one-sigma) random errors were estimated to be typically 4-5 m/s and 3 m/s for Mie-cloudy HLOS winds, but with high variability depending on the signal levels which vary with meteorological conditions. The magnitude of the Mie HLOS wind noise is close to meeting the mission requirements in the free troposphere; however, the Rayleigh noise is larger than the pre-launch mission requirements. The systematic errors (biases) are complex and vary with time. It was necessary to use a bias correction scheme with the ECMWF model as a reference in OSEs to provide a significant positive impact from Aeolus. An understanding of the dominant sources of bias was found via relating O-B departures to the satellite’s housekeeping datasets — particularly for the Rayleigh biases which were found to depend strongly on the temperature of the instrument’s main telescope. A bias correction scheme using the instrument temperatures as predictors was developed as part of the L2B processing, removing the need to bias correct Aeolus winds in ECMWF’s data assimilation system.
OSEs were done using the ECMWF data assimilation system for three periods of the mission. The impact of the assimilation of Aeolus L2B HLOS winds on short range forecasts is demonstrated to be positive, via statistically significant improvements in the forecast fit to other observation types sensitive to temperature, wind and humidity (such as radiosonde observations, GNSS radio occultations, aircraft observations and humidity sensitive microwave radiance observations). The largest short-range impact is found in the tropical upper troposphere (at ~150 hPa); however positive impact can be seen from surface to ~35 km altitude. The forecast impact is positive in the tropical upper troposphere and lower stratosphere and in polar regions; particularly at 2-3 day forecast range; reaching ~2% improvement in RMS error for wind and temperature. Most of the tropical impact comes from the Rayleigh winds, but the Mie provides more of the impact near the poles. It has been shown that the impact of Mie winds can be improved by accounting for representativeness error.
Aeolus’ impact on forecast skill is very good given the higher than pre-launch expected noise levels, complex biases and despite the relatively small size of the assimilated Aeolus dataset compared to most other components of the observing system. The forecast sensitivity observation impact (FSOI) metric, that is available since Aeolus went operational at ECMWF on 9 January 2020, also confirms that Aeolus provides a useful contribution to the global observing system.
The European Space Agency’s Aeolus mission has been demonstrated to be a success as the first space-based Doppler Wind Lidar mission, by providing wind observations of good enough quality to improve weather forecasting. This conclusion was reached by comparing the Level-2B (L2B) horizontal line-of-sight (HLOS) wind observations to the ECMWF NWP model equivalents and by the positive impact of Aeolus in Observing System Experiments (OSEs).
The L2B Rayleigh-clear HLOS wind (one-sigma) random errors were estimated to be typically 4-5 m/s and 3 m/s for Mie-cloudy HLOS winds, but with high variability depending on the signal levels which vary with meteorological conditions. The magnitude of the Mie HLOS wind noise is close to meeting the mission requirements in the free troposphere; however, the Rayleigh noise is larger than the pre-launch mission requirements. The systematic errors (biases) are complex and vary with time. It was necessary to use a bias correction scheme with the ECMWF model as a reference in OSEs to provide a significant positive impact from Aeolus. An understanding of the dominant sources of bias was found via relating O-B departures to the satellite’s housekeeping datasets — particularly for the Rayleigh biases which were found to depend strongly on the temperature of the instrument’s main telescope. A bias correction scheme using the instrument temperatures as predictors was developed as part of the L2B processing, removing the need to bias correct Aeolus winds in ECMWF’s data assimilation system.
OSEs were done using the ECMWF data assimilation system for three periods of the mission. The impact of the assimilation of Aeolus L2B HLOS winds on short range forecasts is demonstrated to be positive, via statistically significant improvements in the forecast fit to other observation types sensitive to temperature, wind and humidity (such as radiosonde observations, GNSS radio occultations, aircraft observations and humidity sensitive microwave radiance observations). The largest short-range impact is found in the tropical upper troposphere (at ~150 hPa); however positive impact can be seen from surface to ~35 km altitude. The forecast impact is positive in the tropical upper troposphere and lower stratosphere and in polar regions; particularly at 2-3 day forecast range; reaching ~2% improvement in RMS error for wind and temperature. Most of the tropical impact comes from the Rayleigh winds, but the Mie provides more of the impact near the poles. It has been shown that the impact of Mie winds can be improved by accounting for representativeness error.
Aeolus’ impact on forecast skill is very good given the higher than pre-launch expected noise levels, complex biases and despite the relatively small size of the assimilated Aeolus dataset compared to most other components of the observing system. The forecast sensitivity observation impact (FSOI) metric, that is available since Aeolus went operational at ECMWF on 9 January 2020, also confirms that Aeolus provides a useful contribution to the global observing system.