Sensitivity of MJO to the CAPE lapse time in the NCAR CAM3.1

Ping Liu, Bin Wang

International Pacific Research CenterUniversity of Hawaii

Sponsored by SciDAC project, computations partly finished at SDSCThanks to: Jerry Meehl

2007 CCSM AMWG meeting at NCAR

In CAM3.1 T42L26, MJO is weak in amplitude and irregular in propagation as in CCM3 (Maloney 2001) and CAM2 (Liu et al 2005)

Variance of 20-80-day filtered U850 inextended winter season (NDJFMA) during 1979-2001

Variance of 20-80-day filtered precipitation inextended winter season (NDJFMA) during 1979-2001

Variance of 20-80-day filtered OLR inextended winter season (NDJFMA) during 1979-2001

Power spectra(10N-10S)1979-2001Winter Nov-Apr850hPa u

Power spectra(10N-10S)1979-2001Winter Nov-AprOLR

IrregularRegression of U850

Onto 155E inExtended winter

During 1979-2001With filtered data

Why?• Observational studies indicate a close coupling

exists between large-scale disturbances and convection associated with MJO (Wang 1988, …)

• A precondition of moisture (or buildup) by boundary layer convergence and/or shallow convection before deep convection associated with MJO bursts (Hendon, Salby, Maloney, Sperber…)

Why?

• Experiments with CCM3 (Maloney 2001, Zhang 2005) and CAM2 (Liu 2005) disclose that either model with alternative convective schemes or a revised closure can simulate much improved MJO although deficiencies remain

• Consequently the convective schemes probably have flaws in 1) deep convection configuration; 2) partition of deep/shallow convection

Where?Basic theories in the Zhang and McFarelane

(1995) scheme for deep convection

(1) A mass flux scheme based on Quasi-Equilibrium

theory (Arakawa and Schubert 1974)

(2) Uniform mass flux at cloud base for updraft

(3) Convection is triggered wherever there is net

positive CAPE (including CIN). Or CAPE

threshold is positive (70 J/kg in code).

(4) Scheme closed on CAPE consumed exponentially

at a specified time scale (2 hours in paper, 1 hour

in code: tau=3600.).

Hypothesis“Convection frequently occurs pre-maturely in the CCSM2”

(Dai 2004). Add a RH threshold for triggering deep convection can enhance the precipitation variability (Zhang and Mu 2005) but not for the RAS (Maloney 2001) in CCM3. A too frequent deep convection might prevent a reasonabe partition of shallow/deep convection then moisture buildup does not occur. So

(3) Is the CAPE threshold low?

The QE theory requires

(4) the specified time for CAPE lapse too short?

s

sADJLS

5LS

43ADJ

10~ lfor typica AS74) in 154 (EQ

hours 10 ~ )1010(~ ,

More evidence“Tropical atmosphere have a thermal-dynamical

background of CAPE at 1000 J/kg” – Heat engine theory by Renno (1996; reversible)

Zhang and McFarelane (1995) table

LTM (1979-2001) DJFM Pseudo-Adiabatic CAPE, J/kg interval 500, thick 1500

Experiments

CAPE threshold: 3 and 10 times

CAPE lapse time: 1, 2, 4, 6, 8, 10 hours

Run: AMIP 1978.9 ~ 2002.8

Model: CAM3.1 T42L26

Results

CAPE threshold: 3 and 10 times

210, 700 J/kg

LTM (1979-2001) DJFM P-A (left) and RV (right) CAPE, J/kg

LTM (1979-2001) DJFM precipitation, mm/dayInterval 3, thick 9

Regression of U850Onto 155E in

Extended winterDuring 1979-2001With filtered data

Indications from CAPE threshold experiments

Lifting the CAPE threshold to 10 times as large as that in control can help the mean state and MJO to some extent, but obviously cannot significantly improve the structure of MJO.

Results

CAPE lapse time: 4, 6, 8, 10 hours

LTM (1979-2001) DJFM Reversible CAPE, J/kg interval 200, thick 800

LTM (1979-2001) DJFM Pseudo-Adiabatic CAPE, J/kg interval 500, thick 1500

LTM (1979-2001) DJFM precipitation, mm/dayInterval 3, thick 9

LTM (1979-2001) DJFM Reversible CAPE, J/kg interval 200, thick 800

Power spectra(10N-10S)1979-2001Winter Nov-Mar850hPa u

Power spectra(10N-10S)1979-2001Winter Nov-MarOLR

MJO based on ZM8HR

Variance of 20-80-day filtered U850 inextended winter season (NDJFMA) during 1979-2001

Variance of 20-80-day filtered precipitation inextended winter season (NDJFMA) during 1979-2001

Variance of 20-80-day filtered precipitation inextended winter season (NDJFMA) during 1979-2001

Regression of U850Onto 155E in

Extended winterDuring 1979-2001With filtered data

Frictional convergence in a composite MJO life cycle

• Maloney (1998; 2001)

• Liu (2005)

NOAA

ZM8HRFirst two EOFs of10N~10S mean filtered OLR

Power spectra(10N-10S)1979-2001Winter Nov-MarOLR

Power spectra(10N-10S)1979-2001Winter Nov-MarOLR

Partitioning of shallow and deep convection

LTM (1979-2001) DJFM ratio of shallow convective to total precipitationInterval 10%, shaded >= 50%

LTM (1979-2001) DJFM ratio of shallow (left) and deep (right) convective to total precipitation

Interval 10%, shaded >= 50%

Summary

. Lifting the CAPE threshold does not significantly improve MJO

. Lengthening the CAPE lapse time enhances MJO variability and improves its structure.

. The CAPE lapse time is optimal at 8 hours to simulate the MJO in both variability and structure. Frictional convergence mechanism functions from the Indian Ocean to western Pacific, which is close to observational facts. A 4:5 partition of shallow and deep convection is a key feature in this case.

. ZM8HR is an ideal starting point for further work.

Future work

Local CAPE lapse time

Standard plots for ZM8HR based on AMWG packages 1 and 2, please see

http://www.soest.hawaii.edu/pliu/zm8hr/1/sets.htm

http://www.soest.hawaii.edu/pliu/zm8hr/2/variab.html