1. On the mechanism of eastward-propagation of super cloud clusters (SCCs) over the equator – Impact of precipitation activities on climate of East Asia – ICMCS-V_061101 Masanori YOSHIZAKI and Tomoe NASUNO (IORGC/JAMSTEC) Topics 1. Simple model: linear, 4-layer model with constant N and no basic wind 2. Extension of simple model using a NICAM output (Diabatic heating: positive-only wave CISK) Thanks to Drs. T. Nasuno and M. Sato for providing NICAM data

2. History and motivations ・ Hayashi ･ Sumi (1986) found an eastward- propagating mode around the equator in the aqua-planet numerical experiment. ・ Eastward-propagating super cloud clusters （ SCCs) were obtained by satellite data, too. (e.g., Nakazawa ， Murakami ， Takayabu et al.) Many theories to explain the mechanisms of eastward-propagating modes: 1) Atmospheric instability ・ Moisture convergence ・ Surface evaporation 2) Atmospheric response to independent forcing ・ Tropical intraseasonal stationary forcing ・ Tropical stochastic forcing ・ Lateral forcing Zhang (2003) day East Westward propagating Eastward propagating Nakazawa Which mechanisms are working? Atmospheric instability, or atmospheric response to independent forcing? Intraseasonal variation >>> MJO (Madden-Julian Oscillation)

3. ・ Atmosphere with constant N and no basic wind, ・ Equatorial-beta plane system (β E ), ・ 4 layers in the vertical, ・ Linear system, ・ Heating: positive-only wave-CISK, ・ Large second-order horizontal diffusion. Yoshizaki (1991a,1991b): a simple model of S ＣＣ s 0 ：ｗ B < 0 Q= ｗ B ・ f(z) ：ｗ B > 0 Model: * Horizontal direction: Grid * Vertical direction: Mode expansion

4. Height ｗBｗB botto m Model top Q /N 2 10 Total Q Each modes of Q * Vertical mode expansion Height Combination of two modes Only 1 st mode ｗBｗB * Two heating profiles were considered: * Top-heavy heating profile can be expressed as a combination of positive 1 st mode and negative 2 nd mode η 1 = 1.5, η 2 =0.0 η 1 =1.5, η 2 =-1.5

5. Height ｗBｗB η 1 = 1.5, η 2 =0.0 η 1 =1.5, η 2 =-1.5 Time along the equator Convective mode moves westward in β E. Eastward-propagating mode grows faster than westward- propagating mode in β E. → Changes of vertical heating profiles induce different characteristic features of propagation!

6. * In this model, it is assumed that diabatic heating is greater than adibatic cooling due to upward motion in some layers. However, is the ‘>’ case right, observationally or numerically?; Disturbance driven by convection for the ‘>’ case, or neutral wave in the stable stratification for the ‘<‘ case. Further study could not be pursued in 1990’s, however, because there was no step to check above-mentioned features. Recently, numerical outputs using a global NH model (NICAM) were available. >>> Which mechanisms are working? Atmospheric instability, or atmospheric response to independent forcing?

7. Snapshot of ‘NICAM’ precipitation - Aqua planet - NICAM: Nonhydrostatic ICosahedral Atmospheric Model = Global cloud-resolving nonhydrostatic model

8. 40000 km / 30 days ～ 15.4 m / s SCC 7 km resolution 2S – 2N average x - t distribution of diabatic heating

10. (1) Comparison of Q (diabatic heating) and adiabatic cooling due to upward motion

11. (2) Comparison of Q (diabatic heating) and adiabatic cooling due to upward motion Disturbances driven by convection ( or atmospheric instability ) D is positive in some layers in the vertical direction.

12. Governing equations * 54 vertical grid model is used. * Parameter ε: 0 or 1 ε 1 : Linear or nonlinear (NL) ε 2 : With or without basic eastward wind ε 3 : Including or excluding Rayleigh damping (function of z) Positive only wave-CISK

13. X - time section of vertical motions at the height of 3.7 km along the equator Blue ： upward motion Red : downward motion Full model: ε 1 = ε 2 = ε 3 = 1 η=60 Time (day) X (10,000 km) Vertical section Horizontal section 16 days ~ 29 m / s Yoshizaki (1991a,1991b) * Linear * No zonal wind * Constant N * 4 layers in the vertical * Mode expansion in the vertical * Combination of 1 st and 2 nd modes Present calculation * Nonlinear * Zonal wind * Variable N * 54 layers in the vertical * Grid in the vertical * Diabatic heating simulated by NICAM

14. θ Heating Full model : Basic wind u + N + positive-only wave-CISK + Rayleigh damping Rayleigh damping is working well. Z X Vertical pattern of simulated SCC ｓ

15. Conclusions 1) Diabatic heating is larger than adiabatic cooling due to upward motion in some vertical layers: SCCs appeared in NICAM is disturbances driven by convection. Then, SCCs are excited due to atmospheric instability. 2) The simple model is extended using the NICAM output. 3) When positive-only wave-CISK is applied as diabatic heating, eastward-propagating disturbances appear as a dominant mode. 4) MJO (or SCC) is responsible for the formation of tropical cyclones affecting East Asia. Thus, this study is important. Further studies 1) This vertical grid model should extend to a vertical mode model, to confirm results obtained by a simple vertical mode. 2) Rayleigh damping is important to eliminate the reflection of vertically propagating gravity waves. The differences between inclusion/exclusion of Rayleigh damping should be studied. 3) Multi-scale horizontal feature is not simulated due to selection rule of convection.

19. Horizontal pattern of simulated SCC ｓ

20. Height Case of two vertical modes Case of one vertical mode ｗBｗB Two heating profiles were considered: * Top-heavy heating profile can be expressed as a combination of positive 1 st mode and negative 2 nd mode η 1 = 1.5, η 2 =0.0 η 1 =1.5, η 2 =-1.5 Time along the equator → Changes of vertical profiles of heating induce different characteristic features of propagation!

21. Why does the difference of heating profiles produce different features? Case of one vertical mode (η 1 >1) Only convective mode excited X along the equator η 1 = 1.5, η 2 =0.0 Similarly to an usual convection, disturbances with no propagation are excited Convective mode grows without propagation in no β E. Convective mode moves westward in β E.

22. Why does the difference of heating profiles produce different features? Case of two vertical modes (η 1 >1, η 2 <0) Convective and oscillation modes excited simultaneously Oscillation mode can be separated into EP and WP modes. EP mode grows faster than WP mode in β E.

23. Growth rate Horizontal wavenumber Small Large Horizontal diffusion = 0 Growth rate Horizontal wavenumber Small Large Small horizontal diffusion Growth rate Horizontal wavenumber Small Large Large horizontal diffusion Selection rule of convection In the linear atmosphere system, there are two independent modes; (1) neutral wave modes and (2) exponentially growing modes. (1) Gravity wave, Kelvin wave, Rossby wave and so on: When forced, a selection rule does not work: All waves stimulated by forcing are evenly excited and appear following a dispersion relation. (2) Baroclinic waves, Benard convection, shear instability and so on: Modes with maximum growth rate grow fastest and a selection rule works. In this model, a positive-only wave-CISK works like usual convection and disturbances with maximum growth rate are infinitesimally small without horizontal diffusion (and viscosity). >>>> A large horizontal diffusion is included to get modes with horizontal scales of 1000 km. >>>> No multi-scale horizontal structure!

24. X - time section of vertical motions at the height of 3.7 km along the equator Blue ： upward motion Red : downward motion Full model: ε 1 = ε 2 = ε 3 = 1 η=60 Time (day) X (10,000 km)

25. Governing equations Parameter ε: 0 or 1 *ε 1 : Linear or nonlinear (NL) *ε 2 : With or without basic eastward wind *ε 3 : Including or excluding Rayleigh damping (function of z)

26. Height Case of two vertical modes Case of one vertical mode ｗBｗB η 1 = 1.5, η 2 =0.0 η 1 =1.5, η 2 =-1.5 Time along the equator → Changes of vertical profiles of heating induce different characteristic features of propagation! Convective mode moves westward in β E. Eastward-propagating mode grows faster than westward- propagating mode in β E.