tropical squall lines as convectively coupled gravity waves: why do most systems travel westward?...

Tropical squall lines as convectively coupled gravity waves: Why do most

systems travel westward?

Stefan Tulich1 and George Kiladis2

1CIRES, University of Colorado, Boulder CO, USA

2NOAA ESRL, Boulder CO, USA

Funding: NSF ATM-0806553

Objectives

1) Provide evidence that many tropical “squall line systems” are part of a broad family of disturbances that arise through coupling between convection and tropospheric gravity waves

2) Start to address the question of why most of these wave disturbances move westward

Outline

1) Brief historical review of tropical squall lines - how did we come to know about them; current state of knowledge

2) Analysis of observational data - provide evidence to support the idea

3) Explicit simulations of convection on an equatorial beta-plane - test hypothesis about what causes westward bias

4) Conclusions and future work

Historical Review of Tropical Squall Lines

If one goes back to the earliest papers by leading authors, they’ll be pointed to two even earlier papers on west African squall lines

West African “Disturbance Lines”

• Hamilton and Archibald (1945; QJRMS; No previous articles referenced!) • Eldridge (1957; QJRMS; 2 articles referenced)

West African “Disturbance Lines”

• Hamilton and Archibald (1945; QJRMS; No previous articles referenced!) • Eldridge (1957; QJRMS; 2 articles referenced)

25 deg / 45 hr = 17 m/s

The Thunderstorm Project (1947; USA)

Newton (1950; J. Meteor.) “Structure and mechanisms of the prefrontal squall line”

The Thunderstorm Project (1947; USA)

Newton (1950; J. Meteor.) “Structure and mechanisms of the prefrontal squall line”

The Line Islands Exp. (1967 Cntrl. Pac.)

Zipser (1969; J. Appl. Meteor.) “The role of organized unsaturated downdrafts in the structure and decay of an equatorial disturbance”

15 m/s

The Line Islands Exp. (1967 Cntrl. Pac.)

Zipser (1969; J. Appl. Meteor.) “The role of organized unsaturated downdrafts in the structure and decay of an equatorial disturbance”

GATE (1974; Eastern Atlantic)

• Several squall lines sampled as they passed across the IFA

• Barnes and Sieckman (1984; MWR) “The environment of fast- and slow-moving tropical mesoscale convective cloud lines”

GATE (1974; Eastern Atlantic)

• A number of squall lines sampled as they passed across the IFA

• Barnes and Sieckman (1984; MWR) “The environment of fast- and slow-moving tropical mesoscale convective cloud lines”

Vn > 7 m/s Vn < 3 m/s

TOGA-COARE (1992; Eq. west Pac.)

• Similar to GATE but satellite data more accessible

• Linear MCS-scale bands dominate total rainfall• Numerous fast-moving “2-day waves” were sampled

TOGA-COARE (1992; Eq. west Pac.)

2-day wave composite evolution

Haertel and Johnson (1998)

TOGA-COARE (1992; Eq. west Pac.)

2-day wave composite evolution

Haertel and Johnson (1998)

~ 1500 km

TOGA-COARE (1992; Eq. west Pac.)

2-day wave composite evolution

Haertel and Johnson (1998)

16 m/s

TOGA-COARE (1992; Eq. west Pac.)

Takayabu et al. (1996)

2-day wave vertical cloud evolution

TOGA-COARE (1992; Eq. west Pac.)

Takayabu et al. (1996)

2-day wave vertical cloud evolution

Are 2-day waves just large-scale squall lines?

TOGA-COARE (1992; Eq. west Pac.)

Takayabu et al. (1996)

2-day wave vertical cloud evolution

Are 2-day waves just large-scale squall lines?

Or are squall-lines mini-versions of 2-day waves?

Observational Analysis

• Goal: Advance the idea that many tropical squall line systems are part of a broader family of convectively coupled gravity wave disturbances

• Strategy: Space-time spectral (Fourier) analysis of high-resolution satellite data

Space-time spectral analysis: Previous work

Wheeler and Kiladis (1999)

Power Spectrum of OLR (symmetric component)

Westward Eastward

96 days

3 days

-15 15

1.25 days

Space-time spectral analysis: Previous work

Wheeler and Kiladis (1999)

Power Spectrum of OLR (symmetric component)

Space-time spectral analysis: Previous work

Wheeler and Kiladis (1999)

Power Spectrum of OLR (symmetric component)

Kelvin waves (3-10 day) Eq. Rossby waves(6-50 day)

Westward inertia-gravitywaves (1.3-2.5 day)

Spectral Analysis of TRMM• TRMM 3B42 Rainfall Product

• 1) Global from 50N-50S• 2) 0.25 deg. resolution in space• 3) 3-hourly in time (1999-present)

TRMM TMI CPC Global Merged IR

Spectral Analysis of TRMM• TRMM 3B42 Rainfall Product

• 1) Global from 50N-50S• 2) 0.25 deg. resolution in space• 3) 3-hourly in time (1999-present)

TRMM rainfall spectrum

96 days

3 days

1.7 days

Looking at smaller scales

96 days

12 hrs

1 day

Looking at smaller scales

Sharp diurnal peak

96 days

12 hrs

1 day

Looking at smaller scales

Sharp diurnal peak

hn ~ 20-40 m

96 days

12 hrs

1 day

Looking at smaller scales

Sharp diurnal peak

cn ~ 14-20 m/s

96 days

12 hrs

1 day

Looking at even smaller scales

96 days

6 hrs

12 hrs

Looking at even smaller scales

~ 6-hr periods &~ 400-km wavelengths

96 days

6 hrs

12 hrs

Where are these signals most active?

“WIG” filter window

96 days

6 hrs

12 hrs

Map of WIG-filtered variance (Boreal Summer JJA)

Focus on N. Africa (JJA)

Focus on N. Africa (JJA)

Hovmollers of rainfall over N. Africa (7.5-12.5N)

2005 2006 2007

Hovmollers of rain over N. Africa (7.5-12.5N)

2005 2006 2007

How do these systems relate to objectively identified squall lines?

AMMA 2006 Field Experiment (ROP: July 5 – Sept 27)

Analysis of Niamey Radar Data

Rickenbach et al. (2009; JGR) “Radar-observed squall line propagation…”

Rain Hovmoller + Radar Identified Squall Lines

Linear convective bands during TOGA COARE?

Rickenbach and Rutledge (1998)

Linear convective bands during TOGA COARE?

Rickenbach and Rutledge (1998)

Hovmoller of CLAUS Tb during TOGA COARE (Cruises 2 and 3)

Hovmoller of CLAUS Tb during TOGA COARE (Cruises 2 and 3)

Inclusion of EIG-filtered rainfall

Inclusion of EIG-filtered rainfall

What is the typical evolution of these disturbances?

Strategy:

Lagged linear regression of WIG-filtered rainfall to construct statistical composites

Location of base point

Base point (2E, 10N)

Composite WIG rain evolution (2E,10N)

Note: data averaged between 7.5-12.5 N

Composite WIG rain evolution (2E,10N)

18 m/s

Note: data averaged between 7.5-12.5 N

Composite WIG rain evolution (2E,10N)

18 m/s

~2 day period

Note: data averaged between 7.5-12.5 N

Composite WIG rain evolutionPlan views at lags: -12,0,12 hr

+12 hr 0 hr -12 hr

Comparison to the west Pac.

Composite WIG wave evolution (155E, 5N)

Note: data averaged between 2.5-7.5 N

Composite WIG wave evolution (155E, 5N)

18 m/s

Note: data averaged between 2.5-7.5 N

Composite WIG wave evolution (155E, 5N)

18 m/s

~2 day period

Note: data averaged between 2.5-7.5 N

Side by side comparison

West Pacific West Africa

Side by side comparison

West Pacific West Africa

Side by side comparison

West Pacific West Africa

Side by side comparison (Plan view at lag 0)

West Pacific West Africa

Side by side comparison (Plan view at lag 0)

West Pacific West Africa

Side by side comparison (Plan view at lag 0)

West Pacific West Africa

Oceanic WIG waves as traveling “V”s or “U”s

West Pacific

Takayabu (1994)

Oceanic WIG waves as traveling “V”s or “U”s

West Pacific

Takayabu (1994)

And squall lines too!

West PacificZipser (1969)

Conclusions thus far

• Tropical squall line systems and linear MCSs appear to be associated (if not synonymous) with convectively coupled gravity wave disturbances

• Westward-moving waves dominate, especially over Africa

Idealized numerical experiment

• Explicit, nested simulations of convection on an equatorial beta-plane

• Two types of runs:

1) Zonal-mean u-wind relaxed to zero

2) Zonal-mean u-wind relaxed to shear profile

Idealized numerical experiment

• Explicit, nested simulations of convection on an equatorial beta-plane

• Two types of runs:

1) Zonal-mean u-wind relaxed to zero

2) Zonal-mean u-wind relaxed to shear profile

Idealized numerical experiment

• Explicit, nested simulations of convection on an equatorial beta-plane

• Two types of runs:

1) Zonal-mean u-wind relaxed to zero

2) Zonal-mean u-wind relaxed to shear profile

Further details

• Model: WRF (most recent version)

• Forcing: Spatially uniform radiative-like cooling to drive deep convection

• SST: Zonally uniform; peaked at eq.

Further details

• Model: WRF (most recent version)

• Forcing: Spatially uniform radiative-like cooling to drive deep convection

• SST: Zonally uniform; peaked at eq.

Nesting strategy: 3 grids

dx, dy = 27 km

8000 km

9900

km

Equator

45 N

45 S

Grid 1

Nesting strategy: 3 grids

dx, dy = 27 km

8000 km

9900

km

Equator

45 N

45 S

PeriodicPeriodic

Grid 1

Nesting strategy: 3 grids

dx, dy = 27 km

8000 km

9900

km

Equator

45 N

45 S

PeriodicPeriodic

Rigid wall

Rigid wall

Grid 1

Nesting strategy: 3 grids

dx, dy = 9 km

8000 km

15 N

15 S

45 N

45 S

3300

km Grid 2

Nesting strategy: 3 grids

8000 km

15 N

15 S

PeriodicPeriodic

45 N

45 S

Grid 2

dx, dy = 9 km

Nesting strategy: 3 grids

8000 km

15 N

15 S

PeriodicPeriodic

45 N

45 S

dx, dy = 9 km

Grid 2

Nesting strategy: 3 grids

8000 km

15 N

15 S

45 N

45 S

5 N

5 S

Grid 3dx, dy = 3 km

• Coriolis force acts only on perturbation winds (about the zonal mean)

• Prevents the formation of unwanted zonal jets and tradewinds

One last detail

Results

Rain hovmoller: No shear

Rain hovmoller: No shear

Rain spectrum: No shear

Rain hovmoller: Shear

Rain hovmoller: Shear

Rain spectrum: Shear

Other shear profiles

Hovmoller for shear reversal

Conclusions

• Vertical shear of background zonal wind is essential for producing westward bias in convective wave propagation

• Simulated “V”-pattern in cloudiness consistent with observations of oceanic squall lines and 2-day waves

Implications of “V” pattern

Radar 1

Radar 2

Implications of “V” pattern

Radar 1

Radar 2

Implications of “V” pattern

Radar 1

Radar 2

Implications of “V” pattern

Radar 1

Radar 2

Implications of “V” pattern

Radar 1

Radar 2

Fast-mover;Shear perpendicular

Slow-mover;Shear parallel

Open Questions

• Why are two-day periodicities absent from the model?

• Why is low-level shear important?• Role of topography/diurnal forcing?• What determines the “V” vs. N-S line

structure?• Implications of westward bias towards the

QBO?

What about the squall lines observed during GATE?

Going back to the first geostationary satellite IR dataset

(SMS-1; Smith & Vonderhaar 1976, CSU Tech note.)

• hourly at ~ 0.1 deg

What about the squall lines observed during GATE?

Going back to the first geostationary satellite IR dataset

(SMS-1; Smith & Vonderhaar 1976, CSU Tech note.)

What about the squall lines observed during GATE?

Going back to the first geostationary satellite IR dataset

(SMS-1; Smith & Vonderhaar 1976, CSU Tech note.)

Typical (18-day window) power spectrum of SMS Tb observed during GATE

Hovmoller of SMS Tb (<250 K) during GATE

Squall line dates reported by Houze and Rappaport (1984)

Hovmoller of CLAUS Tb during TOGA COARE (Cruises 1 and 2)

Hovmoller of CLAUS Tb during TOGA COARE (Cruises 1 and 2)

1

Hovmoller of CLAUS Tb during TOGA COARE (Cruises 1 and 2)