Mesoscale prediction in the Antarctic using cycled ensemble data assimilation

View More View Less
  • 1 School of Meteorology, University of Oklahoma, Norman, Oklahoma
  • 2 School of Meteorology, University of Oklahoma, Norman, Oklahoma
  • 3 School of Meteorology, University of Oklahoma, Norman, Oklahoma
© Get Permissions
Restricted access

Abstract

Due in part to sparse conventional observation coverage in the Antarctic region, atmospheric studies in this part of the globe often rely more heavily on numerical models. Model representation of atmospheric processes in the Antarctic remains inferior to representation in the Northern Hemisphere midlatitudes. Poor representation may be related to inaccurate model analyses that do not optimally utilize the limited observation network. Here, the ensemble Kalman filter (EnKF) data assimilation (DA) technique is employed in lieu of variational DA techniques to investigate impacts on model analysis accuracy. This DA technique (provided by Dart Assimilation Research Testbed; DART) is coupled with a polar-modified, mesoscale numerical model which together comprise Antarctic-DART (A-DART). A-DART is cycled with DA and run over a one-month period, assimilating only conventional observations. Results show relatively good agreement between A-DART and observations. Comparison with radiosonde temperature and geostationary satellite wind observations shows large differences between RMSE and ensemble spread in the upper troposphere. The analysis increment shows large values in the eastern Atlantic–western Indian Oceans associated with geostationary satellite wind observations. Further evaluation determines that geostationary satellite wind observations may be biased in this region. Overall, this baseline demonstration of ensemble-based modeling applied in the Antarctic produced short-term forecasts which were competitive with two operational modeling systems while assimilating on the O(106) fewer observations. A-DART is capable of assimilating additional observations for a variety of applications. This study highlights the capability of applying this ensemble-based DA technique for process and forecast studies in an observation-sparse region.

Corresponding author address: Christopher Riedel, School of Meteorology, University of Oklahoma, 120 David L. Boren Blvd. Suite 5900, Norman, OK 73072-7307. E-mail: christopher.p.riedel-1@ou.edu

Abstract

Due in part to sparse conventional observation coverage in the Antarctic region, atmospheric studies in this part of the globe often rely more heavily on numerical models. Model representation of atmospheric processes in the Antarctic remains inferior to representation in the Northern Hemisphere midlatitudes. Poor representation may be related to inaccurate model analyses that do not optimally utilize the limited observation network. Here, the ensemble Kalman filter (EnKF) data assimilation (DA) technique is employed in lieu of variational DA techniques to investigate impacts on model analysis accuracy. This DA technique (provided by Dart Assimilation Research Testbed; DART) is coupled with a polar-modified, mesoscale numerical model which together comprise Antarctic-DART (A-DART). A-DART is cycled with DA and run over a one-month period, assimilating only conventional observations. Results show relatively good agreement between A-DART and observations. Comparison with radiosonde temperature and geostationary satellite wind observations shows large differences between RMSE and ensemble spread in the upper troposphere. The analysis increment shows large values in the eastern Atlantic–western Indian Oceans associated with geostationary satellite wind observations. Further evaluation determines that geostationary satellite wind observations may be biased in this region. Overall, this baseline demonstration of ensemble-based modeling applied in the Antarctic produced short-term forecasts which were competitive with two operational modeling systems while assimilating on the O(106) fewer observations. A-DART is capable of assimilating additional observations for a variety of applications. This study highlights the capability of applying this ensemble-based DA technique for process and forecast studies in an observation-sparse region.

Corresponding author address: Christopher Riedel, School of Meteorology, University of Oklahoma, 120 David L. Boren Blvd. Suite 5900, Norman, OK 73072-7307. E-mail: christopher.p.riedel-1@ou.edu
Save