|Title||Atmospheric motion vector observations in the ECMWF system: Second year report|
|Series/Collection||EUMETSAT/ECMWF Fellowship Programme Research Reports|
|Authors||Salonen, K, Bormann, N|
|Event Series/Collection||EUMETSAT/ECMWF Fellowship Programme|
|Place of publication||Shinfield Park, Reading|
The report discusses the recent changes in the operational use of the AMVs in the ECMWF system, and the main results from research work carried out during 2012. One of the research highlights of the year was to take part in the 2nd international AMV impact intercomparison exercise which was coordinated by the international winds working group. In total 7 NWP centres reported results from AMV denial experiments performed over two 1.5 month long periods. The ECMWF results indicate that AMVs have a positive impact especially at high levels in the tropics. The results are consistent with results obtained earlier, and similar to results from other contributing NWP centres. The work on situation dependent observation errors for AMVs has continued. The focus has been on investigating the use of the new observation errors. A set of data assimilation experiments with different configurations has been performed. One main goal has been to simplify the current operational first guess check. Based on the results it seems that removing the asymmetric part from the first guess check is possible without degrading the forecast quality. Investigations on using a criterion to reject observations with high error in wind due to error in the height assignment have also been done. The criterion seems to be beneficial but it should not be too tight to avoid rejecting of good quality observations. The overall impression of the results is that the use of situation dependent observation errors, the simplified first guess check and the criterion for the magnitude of the error due to error in height has a positive impact on the forecast. The operational changes include GOES-15 replacing GOES-11, new AMV processing at Eumetsat using the cross correlation method, and active use of AVHRR AMVs from NOAA-15, -16, and -18. Results from monitoring and experimenting with these new data sets are reported.