The Madden-Julian Oscillation in the NCEP GFS AMIP And CFS CMIP Simulations
By Kyong-Hwan Seo, Wanqiu Wang, Jae-Kyung E. Schemm, and Arun Kumar
The effect of air-sea coupling on the MJO is examined using a 21-year CMIP simulation with the fully coupled NCEP Climate Forecast System model (CFS T62) by comparing an AMIP simulation forced by prescribed sea surface temperature from the uncoupled Global Forecast System model (GFS T62). Additional CMIP simulation is performed with a higher horizontal resolution model (CFS T126) to investigate the impact of the model resolution.
Primary Results
1. The interactive air-sea coupling improves the coherence among convection and circulation and other surface fields.
2. Especially, over the Indian Ocean, the simulated MJO in CFS T62 and T126 shows the observed frictional wave-CISK (conditional instability of the second kind) mechanism.
3. In CFS T126, slight improvements appear in the intensity, structure and propagation speed of the MJO.
4. However, all simulations yield weaker MJO signals than the observations and MJO signals tend to be stalled at the propagation barrier at the Maritime continents and far western Pacific. The possible causes of these model deficiencies include (a) Breakup of equatorial convection in the Indian Ocean, (b) Cold SST bias over the eastern Indian Ocean, (c) Weak vertical easterly wind shear and easterly low-level zonal wind bias over the Maritime continents and western Pacific, (d) Model horizontal resolution, and (e) Physical parameterizations problem.
a) Structure, Propagation and Phase Relationships of MJO Signals
|
FIG. 1. Spatial structures of combined empirical orthogonal functions (EOFs) of 20-100 day filtered OLR, u850, and u200. (a) and (b) EOFs 1&2 from observations, (c) and (d) EOFs 1&2 from GFS, (e) and (f) EOFs 1&2 from CFS T62, and (g) and (h) EOFs 1&2 from CFS T126. OLR, u850, and u200 are plotted in black, blue, and red, respectively. The percentage value above each panel is the variance explained by each mode and scaled against the observations. All variables are normalized by the averaged value of global variance (9.06 Wm-2 for OLR, 1.25 ms-1 for u850, and 3.51 ms-1 for u200).
FIG. 2. Lag correlation between PC2 and u850, precipitation rate, OLR, surface downward solar radiation flux, downward latent heat flux, surface temperature, and 1000-hPa moisture convergence for (a) observations, (b) GFS T62, (c) CFS T62, and (d) CFS T126. The correlation of 0.15 is the threshold value for the 95% significance level.
|
www.nws.noaa.gov 
