Outstanding Issues in MJO Simulation and Initiation (NCAR CAM)

1. Outstanding Issues in MJO Simulation and Initiation (NCAR CAM) Eric D. Maloney Colorado State University Department of Atmospheric Science April 13-14, 2009 DYNAMO Meeting Boulder, Colorado

2. Common Comment from the Literature….. Comments like the following are common: “This model was found unable to sustain the eastward propagating, convectively coupled tropical circulation anomalies produced by the MJO when initialized at times the MJO is active. This particular model furthermore is unable to develop an active MJO when initialized prior to such observed development.” (Hendon et al. 2000).

3. Initiation of MJO Convection in the Indian Ocean Presents a Prediction Barrier January 11, 1993 Forecast Skill (Wind Correlations) • Hindcast experiments with ECMWF forecast model • Lack of skill at forecasting Indian Ocean convective initiation provides a barrier to MJO prediction. Agudelo et al. 2007

4. The Preconditioning of the Indian Ocean Troposphere in Advance of MJO Convection is Not Well Simulated • Tropospheric humidity anomalies in this model do not build to same extent as in observations in advance of MJO initiation, and the strong convective event is missed Agudelo et al. 2007

5. Observations Generally Indicate a Notable Preconditioning of the Indian Ocean Atmosphere in Advance of Convection • The troposphere gradually moistens in advance of MJO convection • Plot is generated from reanalysis data, and humidity field is constrained by the model Benedict and Randall 2007

6. Models Increasing Sensitivity of Convection to Free Troposphere Humidity Improves Representation of MJO in A Climate Model (CAM3/RAS) • Top model’s convection very strongly tied to CAPE Increasing sensitivity Observations Hannah and Maloney (2009?)

7. Moist Static Energy Anomalies, Vertical Structure Moist Static Energy Humidity Portion Units: J kg-1

8. What Contributes to Model Preconditioning? (Moist Static Energy Budget) Vertical Integral Surface Latent and Sensible Heat Flux Shortwave Flux Convergence Longwave Flux Convergence Horizontal advection Vertical advection 30-90 Day Anom. Precip LH Maloney (2009)

9. Intraseasonal Wind Speed (QuikSCAT) vs. Precipitation (TRMM) Correlation Correlation (with Mean Boreal Winter Winds) Araligidad (2007)

10. Suppressed Wind Speed Precedes Initiation of MJO in Indian Ocean Precipitation/Wind Wind Speed/Wind Araligidad (2007)

11. Changing the Basic State the Surface Fluxes Feel Can Change MJO Propagation Speed 1) Strong westerly basic state perturbation (~6 m/s): 2) Feel a weaker westerly basic state perturbation (~3 m/s): 3) Aquaplanet easterly 4) Strong easterly basic state correction (~-4 m/s) Realistic SST Distribution

12. Key Questions on MJO Initiation That DYNAMO Might Be Able to Answer • What combination of processes contribute to this buildup (or slow the buildup) of column moist static energy/humidity in advance of the initiation of MJO convection? • Vertical advection (e.g. shallow convection) • Horizontal advection • Surface Heat Fluxes • Contributing buildup of upper ocean heat content • Why is the preconditioning timescale longer for the MJO than other convectively coupled disturbances? • DYNAMO can aid development of convection parameterizations, which presently extinguish convective instability too early, and damp existing MJO events too quickly

13. Extra Slides

14. Propagation of Equatorial 30-90 Day Zonal Wind Anomalies • MJO amplitude is dramatically weakened when the rain re-evaporation fraction is diminished. • While such behavior is often attributed to “moisture-convection feedbacks”, in this model it appears to be due to the impact of rain re-evaporation on the model climatology. NCEP Reanalysis CAM 3 RAS (a=0.6) CAM 3 RAS with Reduced Rain Re-evaporation (a = 0.05)

15. 850 hPa Wind Spectra

16. Variance Ratios Eastward rel. to westward Eastward rel. to observations

17. Variance Ratios in Other Models SPCAM Kim et al. 2009

18. Many Models Produce Indian Ocean Invariability that Is Weaker Than Observed

19. Models Typically Have a Difficult Time Capturing this Buildup Kim et al. 2009

20. Partitioning of Moist Static Energy Budget (Vertical Integral) Vertical Integral Surface Latent and Sensible Heat Flux Shortwave Flux Convergence Longwave Flux Convergence Horizontal advection Vertical advection 30-90 Day Anom. S Precip LH

21. Vertical Distribution of Horizontal Advection Anomalies S Units: 10-4 W kg-1

22. Partitioning of Meridional Component of HADV = 50-day avg. is the deviation from the 50-day avg. Sum

23. Eddy Kinetic Energy Anomalies at Days -20, 0 Time of Peak Moistening, Easterly Anomalies Units: m2 s-2 Time of Peak Drying, Westerly Anomalies

24. Conclusions/Future Work

25. Aquaplanet Simulations

26. Thanks!

27. Extra Slides

28. A Role for Wind-Induced Latent Heat Flux? Units: mm day-1 Units: W m-2

29. Eddy Kinetic Energy Anomalies versus Time S 850 hPa 700 hPa = 10-day avg. is the deviation from the 10-day avg.

30. Observed EKE Shows Similar Variations During MJO Events • In nature, westerly MJO phases are characterized by enhanced EKE, and easterly MJO phases are characterized by suppressed MJO activity • These variations are consistent with the model in phase and magnitude. Maloney and Dickinson (2003) Units: m2 s-2

31. Dominant 850 hPa Eddy Structure in the Model NH Regression Point SH Regression Point Units: 10-6 s-1, Contour interval 2.0

32. Mean 700 hPa Wind, Humidity (December-May) NCEP ERA40 CAM 3 RAS (a=0.6) Units: g kg-1

33. Mean 850 hPa Wind, Humidity (December-May) Units: g kg-1 ERA40 CAM 3 RAS (a=0.6) CAM 3 RAS with Reduced Rain Re-evaporation (a = 0.05)

34. Consistent with Previous Results (Maloney and Hartmann 2001) • Shutting down rain evaporation and downdraft parameterizations in a GCM removed the MJO. • The MJO could be recovered if the temperature and humidity forcing of the downdrafts and rain evaporation were simply applied in a time-invariant sense to reproduce the control climatology. Control

35. Consistent with Previous Results (Maloney and Hartmann 2001) Zonal Mean Q Difference ~3.2 g kg-1 20N Eq 20S 10S 10N • The largest climatology difference in the model with rain evaporation and without was a moister equator with heightened meridional humidity gradient. • The largest differences are between 700 and 850 hPa.