Monday, 23 January 2017
4E (Washington State Convention Center )
Hailan Wang, NASA, Hampton, VA; and D. Achuthavarier, Y. Chang, and S. Schubert
In this study, we investigated the NASA GEOS-5 AGCM bias and assessed its impact on model hindcast of the Madden-Julian Oscillation (MJO). The current GEOS-5 AGCM is deficient in simulating the MJO, including its initiation. To better identify and understand the model deficiency, we have produced at NASA GMAO a global GEOS-5 reanalysis that assimilates global observations including the quality-controlled in-situ observations collected during the DYNAMO field campaign (1October 2011-31March 2012), referred to as the DYNAMO reanalysis. The distribution of the analysis tendency terms therein shows the GEOS-5 AGCM to have cold and dry biases in the lower troposphere, a manifestation of the poor model representation of shallow convection, warm and wet biases in the mid-upper troposphere, and a cold bias at about 150hPa, over the tropical Indian Ocean. These AGCM biases show an overall weak dependence on MJO phases, and persist throughout much of the DYNAMO period. In the DYNAMO reanalysis, the assimilation of DYNAMO observations helps overcome these AGCM biases over the tropical Indian Ocean and improves considerably the reanalysis representation of the observed MJO initiation processes.
Motivated by the above finding, we next investigated the impact of the GEOS-5 AGCM biases on the model hindcast of the MJO. Here the model bias correction was constructed as 6-hourly climatological (2008-15) Incremental Analysis Update (IAU) analysis increments, using the DYNAMO reanalysis for the DYNAMO period and MERRA-2 for the rest periods. To correct the model bias in a MJO hindcast run, the above constructed 6-hourly climatological IAU corrections were applied to model basic variables within the free-running AGCM. Using the standard and the bias corrected GEOS-5 AGCM, two series of daily AGCM hindcast experiments have been performed. The comparison of AGCM hindcast skill between the two experiments shows that the model bias correction improves considerably the MJO hindcast skill beyond the first 10 days when the contribution from initial conditions fades away. The improved hindcast skill is mainly attributable to the improved air temperature and specific humidity, their vertical profiles in particular, which improves the model representation of physical processes (e.g., diabatic heating and moistening) associated with the MJO development.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner