Mechanisms of poleward propagating, intraseasonal convective anomalies in a cloud-system resolving model William Boos & Zhiming Kuang Dept. of Earth & Planetary Sciences Harvard University October 16, 2009
Outline Background and observations Results from quasi-2D models with explicit convection Mechanisms of instability and propagation Main message: For intraseasonal convective anomalies during boreal summer: Poleward propagation occurs due to convectively-coupled beta-drift of a vorticity strip Instability occurs due to moisture-radiation feedback
Boreal summer MJO lifecycle of TRMM precip diagnostic from CLIVAR MJO working group, based on EOFs after Wheeler & Hendon (2004) propagation has prominent poleward component some events do exhibit poleward propagation without eastward propagation
Viewed as poleward migration of ITCZ 1.5 m/s NOAA OLR anomalies, °E, summer 2001 Several events typically occur each boreal summer, modulating intensity of South Asian monsoon
History of axisymmetric model studies Land-atmosphere interactions (Webster & Chou 1980) Poleward gradient of convective instability (Gadgil & Srinivasan 1990) Dynamical coupling of anomalies to baroclinic mean state (Bellon & Sobel 2008, Jiang et al. 2004) … but all of these studies use idealized parameterizations of moist convection, and mode characteristics depend on convective closure
Test in model with explicit convection System for Atmospheric Modeling (SAM, Khairoutdinov & Randall 2003 ) 1 km horizontal resolution Beta-plane, 70°N – 70°S 4 zonal grid points Oceanic lower boundary with prescribed SST precipitation
Model with wider zonal dimension 4 zonal grid points 32 zonal grid points Precipitation snapshots when ITCZ is near 10N: latitude Old domain: 140° meridional x 4 km zonal New domain: 140° meridional x 960 km zonal For computational efficiency, use RAVE methodology of Kuang, Blossey & Bretherton (2005) : 30 km horizontal resolution, RAVE factor 15 Similar results obtained for RAVE factors ranging from 1-15 at 30 km resolution, and for one standard run with 5 km resolution x (km) mm/day
Precipitation in wide-domain model 0.5 m/s mm/day
Zonal mean vertical structure for wide domain m/s
Composite 950 hPa vorticity Zonal mean vorticity satisfies necessary condition for barotropic instability Anomalies form closed cyclone for part of poleward migration, and zonal strip for remainder Suggestive of “ITCZ breakdown” (Ferreira & Schubert 1997) zonal mean vorticity composite relative vorticity latitude
Animation of two events Poleward drift of vorticity patch/strip on β-plane… coupled to moist convection latitude x grid point Shading: 930 hPa relative vorticity Black contours: precipitation
Schematic: propagation mechanism 1.deep ascent creates (barotropically unstable) low-level vortex strip 3. Ekman pumping in vortex strip humidifies free-troposphere poleward of original deep ascent, shifting convection poleward Convectively-coupled beta-drift of vortex strip deep ascent 2. perturbed vortex strip migrates poleward deep ascent vorticity anomaly x y y z
Test mechanism in dry model β-drift biases low-level convergence poleward of free-tropospheric heating applied (constant) thermal forcing surface meridional wind
Surface wind in dry model latitude constant imposed heating x grid point
Looks like unstable moisture mode J/kg MSE tendencies composite moist static energy anomaly
Model tests of instability mechanism mm/day fixed radiative cooling fixed surface heat fluxes control run Precipitation Hovmollers:
Instability mechanism is non-unique Dashed black lines denote latitude of peak moist static energy anomaly Control run Run with fixed radiative cooling
Summary Axisymmetric cloud permitting models fail to produce robust poleward propagating, intraseasonal convective anomalies Meridional “bowling alley” domains O(1000 km) wide do produce such anomalies – Suggested propagation mechanism: convectively-coupled beta-drift of vortex strip – Anomalies destabilized by moisture-radiation feedback – Perhaps slowed and made more coherent by WISHE – Multiple instability mechanisms can operate, with structural changes Future work: – Behavior in wider domains – Validation of mechanism in simpler models
Additional slides
Wide domain permits high amplitude eddies latitude x (10 5 m) g/kg day 0 day 20day 30day 41day 53 composite 930 hPa wind and humidity
Why does the wide domain make a difference? It’s the eddies… J/kg MSE tendencies composite moist static energy anomaly advective componentstotal & zonal mean advection
Propagation speed scaling Plots of precip and v wind for beta 0.75, 1, 2
Observed vertical structure data: ERA-40 Reanalysis, composite of strong poleward events latitude pressure (hPa) Note some similarties to eastward moving MJO
latitude pressure (hPa) Observed vertical structure data: ERA-40 Reanalysis, composite of strong poleward events
Behavior depends on zonal width, not zonal d.o.f. time (days) latitude 5 km resolution with 32 zonal grid points 30 km resolution with 32 zonal grid points
OLR in wide domain model
Vertical structure for wide domain (green line denotes position of peak precip signal used for compositing) m/s
Turn off both WISHE & radiative feedbacks no WISHE or radiative feedbacks control time (days) mm/day Precipitation:
MSE budget for run without WISHE or radiative feedbacks moist static energy anomaly latitude (degrees) pressure (hPa) moist static energy tendencies W m -2 “Convective downdraft instability” J/kg