School of Meteorology (Defense)

Improved Characterization and Prediction of Antarctic Weather Through Ensemble Data Assimilation and Utilization of the CONCORDSAIS Data Set

Chris Riedel

School of Meteorology

03 December 2015, 1:00 PM

National Weather Center, Room 5820
120 David L. Boren Blvd.
University of Oklahoma
Norman, OK

Knowledge of why forecast models exhibit relatively low predictive skill in the Southern Hemisphere is limited. For the last several decades, improvements in forecast skill for the Southern Hemisphere has lagged those seen in the Northern Hemisphere. In 2010, Global Forecast System (GFS) forecasts were skillful out to eight days in the Northern Hemisphere, but only out to seven days in the Southern Hemisphere. Uncertainties in the forecasts are likely due to, at least partially, the analyses incorrectly representing the atmospheric state. Relatively sparse observations over the Southern Ocean add additional weight on numerical models to produce analyses of the atmospheric state, however numerical models contain large errors deriving from their representation of physical processes. These factors make it difficult to rely on data from numerical models to study phenomena such as the rapid warming occurring on the Antarctic Peninsula or to sea ice trends. The Antarctic Mesoscale Prediction System (AMPS), which is a polar modified version of the Weather Research and Forecasting Model (WRF), is the only Southern Hemispheric operational regional model that offers high-resolution forecasts. AMPS employs a three-dimensional variational (3DVAR) data assimilation technique, which may not be ideal in data sparse regions. We hypothesize that an ensemble data assimilation technique produces more accurate analyses compared to those produced by the 3DVAR technique used by AMPS.
Antarctic-Dart (A-Dart) is a coupling of the AMPS model with the Ensemble Adjustment Kalman Filter (EAKF) data assimilation technique provided by the Dart Assimilation Research Testbed. A-Dart was cycled for a one-month period that coincided with the Concordiasi intensive observation period. A central analysis was created, where A-Dart only assimilated the following conventional observations: radiosondes, aircraft communication addressing and reporting system (ACARS), geostationary satellites, and global positioning system (GPS) soundings. After a month of cycling, a model circulation bias was discovered in the upper atmosphere. In order to produce accurate analyses with this ensemble data assimilation approach, the model circulation bias must be corrected. Three experiments were devised to isolate the circulation bias. First, polar orbiting satellite winds were assimilated, but ultimately exerted little impact on the circulation bias. Second, the shortwave radiation scheme was appended to include the variations in ozone that occur over the Antarctic during our test period. However, little improvement was found in the overall circulation bias, but upper-tropospheric and lower-stratospheric biases over the continent were reduced. Lastly, assimilation of atmospheric infrared sounder (AIRS) profiles had the largest effect on correcting the circulation bias. This result confirms the importance of assimilating a high number of observations to produce an accurate analysis in the Southern Hemisphere. Furthermore, the forecast skill for A-Dart was comparable to GFS and was more accurate than the operational AMPS. Overall, A-Dart offers not only a reliable forecast tool but also a reliable research tool for the Southern Hemisphere region.

School of Meteorology (Defense) Seminar Series website