ECMWF and GFS Model Forecast Verification During DYNAMO: Multi-scale Variability in MJO Initiation over the Equatorial Indian Ocean

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 4 February 2014
Hall C3 (The Georgia World Congress Center )
Brandon W. Kerns, Univ. of Miami/RSMAS, Miami, FL; and S. S. Chen

Improving understanding and numerical model predictions of initiation of the Madden-Julian Oscillation (MJO) over the Indian Ocean (IO) was a main objective of the Dynamics of MJO (DYNAMO) field campaign. Observations from DYNAMO have revealed that multi-scale interactions among convective cloud systems, synoptic and large-scale circulation may play an important role in the initiation of MJO events. This study provides a unique and systematic evaluation and verification of the European Centre for Medium-Range Weather Forecasts (ECMWF) and the NOAA Global Forecast System (GFS) forecasts of a strong MJO event observed during DYNAMO in November-December 2011. The model forecasts of zonal wind and precipitation are compared with the ECMWF analysis and Tropical Rainfall Measurement Mission (TRMM) satellite estimates. The ECMWF analysis assimilated in situ observations from DYNAMO. To examine the multi-scale variability both model forecasts and data are decomposed into various spatial scales--basin scale, large/synoptic scale, and mesoscale. Rather than computing a MJO index for verification, a grid-point verification technique using zonal winds focused on the equatorial IO is used. The models' error growth at the various spatial scales is quantified and compared with persistence forecasts. It is found that 1) both models captured the observed properties in the short-term forecasts at 48-72 h, 2) ECMWF model forecasts are superior in the longer-range, 5-15 days, in terms of both convection and circulation of the MJO initiation, and 3) GFS better re-produced the local daily rainfall in the TRMM data; however, ECMWF's overall total "volumetric" equatorial rainfall--the total volume of rain water produced over the Equatorial IO for each day--is more consistent with TRMM. While it is important to forecast local daily rain rate, volumetric rainfall may be a better measure for the overall model water budget. The latter could be an important verification metric for model forecast globally on longer time scales such as the MJO and its downstream teleconnections.