Tropical intraseasonal variability simulated in the NCEP Global

Tropical intraseasonal variability simulated in the NCEP Global Forecast System and Climate Forecast System models Kyong-Hwan Seo*, Wanqiu Wang** and Jae-K. E. Schemm** * Pusan National University Dept of Atmospheric Sciences, Korea ** CPC/NCEP/NOAA, USA NOAA’s 33 rd Climate Diagnostics and Prediction Workshop, Lincoln, Nebraska October 20 -24, 2008

Objectives To investigate the capability for simulating the tropical intraseasonal variability (focusing on the MJO) in a series of atmosphere-ocean coupled and uncoupled simulations using NCEP operational general circulation models. To evaluate the effects of the following factors on the MJO simulation Air-sea coupling Model horizontal resolution Deep convection parameterization Basic state vertical shear Basic state low-level westerlies SST Low-level moisture convergence

Models NCEP Atmospheric Model: NCEP Coupled model: GFS T 62 (AMIP) CFS T 62 (CMIP: GFS T 62 + GFDL MOM 3) NCEP Coupled high resolution run: CFS T 126 (SAS) NCEP Coupled high resolution run with Relaxed Arakawa-Schubert scheme: CFS T 126 RAS Simulation Period: 15 -20 years

OBS GFS T 62 CFS T 62 Power Spectra over equatorial Indian Ocean CFS T 126 RAS -Red line: calculated power spectra -Blue line: background red spectrum -OBS: pronounced 30 -80 day signal GFS T 62, CFS T 126: less significant - CFS T 126 RAS: significant, vigorous power in 30 -80 day range

Leading 2 EOFs of combined OLR, U 200 & U 850 -CFS T 126 RAS: explained variance close to observation among simulations. - Propagation is not revealed in this plot

u 850 prate -OLR LHTFL DSWRF Tsfc fc moist onverg. T b t c C c s p t c & d a t

Factors for the improved MJO simulation (1) Air-sea interaction (2) Model horizontal resolution (3) Basic state vertical shear (4) Basic state low-level westerlies (5) SST (6) Deep convection parameterization (7) Low-level moisture convergence (8) Vertical profile of diabatic heating

Factors for the MJO: (3) Basic-state vertical wind shear Western Pacific -Vertical easterly shear favors the eastward propagating waves (Zhang and Geller, 1994) -CFS T 126 RAS: shows the smallest easterly shear -Background vertical wind shear is not the most important factor

Factors for the MJO: (4) Background low-level wind -Easterly bias acts as a barrier to the eastward propagation of the MJO (Inness and Slingo 2003; Flatau et al. 1997) -CFS T 126 RAS: shows the easterly bias over the Maritime continent and the western pacific -This is not a major factor

Factors for the MJO: (5) SST -A cold SST bias acts to suppress the development of the MJO convection (many references) -CFS T 126 RAS: shows an increased cold bias over the western Pacific Ocean -This is not a major factor

Factors for the MJO: (6) Deep convecton parameterization -CFS T 126 RAS: Active convective activity over the warm pools induce the enhanced lower-level circulation, which in turn helps maintain the MJO convection -The positive feedback between the convection and circulation induces the continued eastward propagation across the Maritime Continent. -This is a major factor

Factors for the MJO: (7) Low-level moisture convergence -Frictional wave-CISK mechanism is the main paradigm for the development and propagation of the MJO -CFS T 126 RAS: shows the strong surface layer moisture convergence both over the Indian Ocean and the western Pacific, which leads enhanced convection by ~2 -5 days, as similar as the observations -The phasing and magnitude of the lower-level moisture convergence are a key factor

Global Circulation Response to the MJO Convection: OBS -200 h. Pa Streamfunction regressed onto PC 1 and PC 2 -Half life cycle -RED: enhanced MJO convection -Blue: suppressed convection -Tropics: anticyclonic couplet at or west of enhanced convection + tropical westerly anom east of enhanced convection: Rossby. Kelvin wave response -PNA-like response -Continued influence to the Americas

Global Circulation Response to the MJO: CFS T 126 -RED: enhanced MJO convection -Blue: suppressed convection -CFS T 126: convection and streamfunction anomalies are weak -No significant suppressed convection over the western Pacific at t=6 & t=12 weaker circulation response

Global Circulation Response to the MJO: CFS T 126 RAS -RED: enhanced MJO convection -Blue: suppressed convection -CFS T 126 RAS: stronger circulation response -Similar pattern to the observation pattern correlation 0. 84 -0. 91 (vs 0. 47 -0. 78 in CFS T 126)

Convectively Coupled Equatorial Waves: OBS -Observed equatorial OLR anomalies -Antisymmetric component: MRG and n=0 EIG connected to each other -Symmetric component: n=1 ER, n=1 WIG/EIG, Kelvin and MJO -Aligned along equivalent depth of 25 m

CFS T 126 -Both coupled runs: no MRG and n=0 EIG signals shown n=1 WIG, n=1 EIG not generated CFS T 126 RAS -Kelvin wave is weaker than the observation -n=1 ER wave produced but with a slower phase speed bias -Models have the equivalent depth of 12 m -Only significant isolated MJO signal appeared in CFS T 126 RAS

Summary MJO - The interactive air-sea coupling greatly improves the coherence between the convection, circulation and other surface fields. - CFS T 126 RAS produces statistically significant spectral peaks in the MJO spectral band the strength of MJO convection and circulation is considerably improved. - Most of all, the MJO convection signal is able to penetrate into the Maritime Continent and western Pacific. - The proper and persistent interaction between the convection and circulation induces the continued eastward propagation across the Maritime Continent. - The improved MJO simulation in CFS T 126 RAS improves the simulation of extratropical circulation anomalies Convectively Coupled Equatorial Waves - ER wave and Kelvin wave are reproduced! - No statistically significant peaks associated with MRG and EIG waves - All models produce too excessive westward synoptic scale disturbances with periods of less than 5 days - Model-generated eastward Kelvin waves are weaker than the observed.

Discussions Several limitations: First, we are not able to specifically determine which processes or parameters of the RAS scheme are most important for producing the enhanced MJO activity. For instance, the questions which components are responsible for the reduced autocorrelation seen in CFS T 126 RAS and whether or not there is an equivalent selfsuppression process in this deep convection scheme (see Lin et al. 2006) can not be answered. Only the final consequences from the interaction of convection and circulation are seen. Second, the vertical diabatic heating profiles associated with convective and stratiform clouds are not available from the current operational setting. In addition, the causes of the excessive high-frequency variability in the space-time spectral analysis can not be addressed. Carefully designed sensitivity test and more flexible implementation of the operational model would be required to resolve these issues. Nonetheless, this study determined the plausible factors for the improved simulation of the MJO.

SAS vs RAS SAS requires that a moist quasi-equilibrium hypothesis is achieved for the cloud ensemble (or the deepest single plume) at each integral. RAS only relaxes thermodynamic state toward equilibrium rather than making instantaneous adjustments to the equilibrium state as in SAS. A brief description of these convection schemes can be found in Das et al. (2002).