Use and Impact of Automated Aircraft Data in a Global 4DVAR Data Assimilation System

Abstract The use of automated aircraft data Aircraft Meteorological Data Relay (AMDAR) and Aircraft Communication Addressing and Reporting System (ACARS) has recently been extended in ECMWF's operational 4DVAR data assimilation system. Herein, a modified data selection procedure is reported on that allows the use of more aircraft profiling data during the aircraft's ascending and descending phase, and more of the most frequent reports at cruise level. It is shown that the accuracy of analyzed jet streams is improved through these changes, as verified against independent (non–real time) aircraft data that had not been used in the experiments. The modifications are shown to have a clear positive impact on the short- and medium-range forecast performance. The revised aircraft usage was implemented operationally in January 2002. The impact in 4DVAR of profiles from American and European automated aircraft in ascending and descending phase has been tested in a data denial impact study, for January and July 2001. This particular impact study was run partly on the request of the WMO/Commission for Basic Systems (CBS) Expert Team on data requirements and the redesign of the global observing system. Their interest is in testing whether a modern data assimilation system (such as 4DVAR) obtains substantial benefit from the aircraft profiles, which sample very irregularly in space and time, given that America and Europe are relatively well covered by radiosondes and wind profilers. The results show a substantial positive impact of the profiling aircraft data on analysis and forecast accuracy. The short-range forecast performance is improved over North America, the North Atlantic, and Europe. In the medium range a clear positive impact is found in the North Atlantic, the European, and Arctic areas in the winter period, and beyond day 6 in the summer period. These results are statistically significant and support the ongoing WMO initiative for further expansion of the AMDAR/ACARS coverage. The results also illustrate the effectiveness of 4DVAR with respect to observations that are irregularly distributed in space and time.