Scale Interactions in Organized Tropical Convection George N. Kiladis Physical Sciences Division ESRL, NOAA George N. Kiladis Physical Sciences Division ESRL, NOAA

Why Study Tropical Convective Variability? Tropical Convection acts as a primary “heat engine” for the atmospheric circulation Variability in tropical convection has global scale impacts over a variety of time scales Convection is coupled to the ocean within the tropics >Sea surface temperature has a strong influence >Atmospheric disturbances influence SST Tropical Convection acts as a primary “heat engine” for the atmospheric circulation Variability in tropical convection has global scale impacts over a variety of time scales Convection is coupled to the ocean within the tropics >Sea surface temperature has a strong influence >Atmospheric disturbances influence SST

OBSERVATIONS OF WAVES WITHIN THE MJO Time–longitude diagram of CLAUS T b (5S–equator), February 1987

The Madden-Julian Oscillation (MJO) Discovered by Rol Madden and Paul Julian at NCAR in 1971 Characterized by an envelope of convection ~10,000 km wide moving eastward at around 5 m/s Most active over regions of high sea surface temperature (> 27  C) Can have a profound impact on the extratropical circulation Is poorly represented in general circulation models, if at all Composed of a variety of higher frequency, smaller scale disturbances Discovered by Rol Madden and Paul Julian at NCAR in 1971 Characterized by an envelope of convection ~10,000 km wide moving eastward at around 5 m/s Most active over regions of high sea surface temperature (> 27  C) Can have a profound impact on the extratropical circulation Is poorly represented in general circulation models, if at all Composed of a variety of higher frequency, smaller scale disturbances

Shallow Water System (Matsuno, 1966)

where is the meridional gradient of f at the eq is the equivalent depth is the gravity wave speed

Theoretical Dispersion Relationships for Shallow Water Modes on Eq.  Plane Frequency Zonal Wavenumber

Theoretical Dispersion Relationships for Shallow Water Modes on Eq.  Plane Kelvin Inertio-Gravity Equatorial Rossby Frequency Zonal Wavenumber

Kelvin Wave Theoretical Structure Wind, Pressure (contours), Divergence, blue negative

Mixed Rossby-Gravity Wave Theoretical Structure Wind, Pressure (contours), Divergence, red negative

Wavenumber-Frequency Spectral Analyis Decompose into Symmetric and Antisymmetric Fields about the Equator Complex Fourier Transform into wavenumber space at each latitude FFT of each wavenumber into frequency space Average the Power for each wavenumber/frequency by latitude Determine “background” spectrum by smoothing raw spectra Divide raw spectra by background spectra to determine signals standing above the background

OLR power spectrum, 15ºS-15ºN, 1979–2001 (Symmetric) from Wheeler and Kiladis, 1999

OLR power spectrum, 15ºS-15ºN, 1979–2001 (Symmetric) from Wheeler and Kiladis, 1999 Eastward Power Westward Power 1.25 Days 96 Days

OLR power spectrum, 15ºS-15ºN, 1979–2001 (Antisymmetric) from Wheeler and Kiladis, 1999

OLR background spectrum, 15ºS-15ºN, 1979–2001 from Wheeler and Kiladis, 1999

from Wheeler and Kiladis, 1999 OLR power spectrum, 1979–2001 (Symmetric)

from Wheeler and Kiladis, 1999 OLR power spectrum, 1979–2001 (Symmetric) Kelvin Westward Inertio-Gravity Equatorial Rossby Madden-Julian Oscillation

from Wheeler and Kiladis, 1999 OLR power spectrum, 1979–2001 (Antisymmetric)

from Wheeler and Kiladis, 1999 OLR power spectrum, 1979–2001 (Antisymmetric) Mixed Rossby- Gravity Eastward Inertio- Gravity

OBSERVATIONS OF KELVIN WAVES AND THE MJO Time–longitude diagram of CLAUS T b (2.5S–7.5N), January–April 1987 Kelvin waves (15 m s -1 ) MJO (5 m s -1 )

OBSERVATIONS OF KELVIN AND MRG WAVES Time–longitude diagram of CLAUS T b (2.5S–7.5N), May 1987

1998 Brightness Temperature 5ºS-5º N

Kelvin Wave Theoretical Structure Wind, Pressure (contours), Divergence, blue negative

OLR power spectrum, 1979–2001 (Symmetric) from Wheeler and Kiladis, 1999

Regression Models Simple Linear Model: y = ax + b where: x= predictor (filtered OLR) y= predictand (OLR, circulation)

OLR and 1000 hPa Flow Regressed against Kelvin-filtered OLR (scaled -20 W m 2 ) at 10  N, 150  W for June-Aug Day 0 Geopotential Height (contours 2 m) Wind (vectors, largest around 5 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

OLR and 1000 hPa Flow Regressed against Kelvin-filtered OLR (scaled -20 W m 2 ) at 10  N, 150  W for June-Aug Day-6 Geopotential Height (contours 2 m) Wind (vectors, largest around 5 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

OLR and 1000 hPa Flow Regressed against Kelvin-filtered OLR (scaled -20 W m 2 ) at 10  N, 150  W for June-Aug Day-5 Geopotential Height (contours 2 m) Wind (vectors, largest around 5 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

OLR and 1000 hPa Flow Regressed against Kelvin-filtered OLR (scaled -20 W m 2 ) at 10  N, 150  W for June-Aug Day-4 Geopotential Height (contours 2 m) Wind (vectors, largest around 5 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

OLR and 1000 hPa Flow Regressed against Kelvin-filtered OLR (scaled -20 W m 2 ) at 10  N, 150  W for June-Aug Day-3 Geopotential Height (contours 2 m) Wind (vectors, largest around 5 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

OLR and 1000 hPa Flow Regressed against Kelvin-filtered OLR (scaled -20 W m 2 ) at 10  N, 150  W for June-Aug Day-2 Geopotential Height (contours 2 m) Wind (vectors, largest around 5 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

OLR and 1000 hPa Flow Regressed against Kelvin-filtered OLR (scaled -20 W m 2 ) at 10  N, 150  W for June-Aug Day-1 Geopotential Height (contours 2 m) Wind (vectors, largest around 5 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

OLR and 1000 hPa Flow Regressed against Kelvin-filtered OLR (scaled -20 W m 2 ) at 10  N, 150  W for June-Aug Day 0 Geopotential Height (contours 2 m) Wind (vectors, largest around 5 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

OLR and 1000 hPa Flow Regressed against Kelvin-filtered OLR (scaled -20 W m 2 ) at 10  N, 150  W for June-Aug Day+1 Geopotential Height (contours 2 m) Wind (vectors, largest around 5 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

OLR and 1000 hPa Flow Regressed against Kelvin-filtered OLR (scaled -20 W m 2 ) at 10  N, 150  W for June-Aug Day+2 Geopotential Height (contours 2 m) Wind (vectors, largest around 5 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

Mixed Rossby-Gravity Wave Theoretical Structure Wind, Pressure (contours), Divergence, red negative

OLR and 850 hPa Flow Regressed against MRG-filtered OLR (scaled -40 W m 2 ) at 7.5  N,  E, Day-1 Streamfunction (contours 2 X 10 5 m 2 s -1 ) Wind (vectors, largest around 2 m s -1 ) OLR (shading starts at +/- 6 W s -2 ), negative blue

CWC W Direction of Motion Temperature Structure of a Dry Kelvin Wave

CWC W Direction of Motion Temperature Structure of a Dry Kelvin Wave

Temperature at Majuro (7  N, 171  E) Regressed against Kelvin-filtered OLR (scaled -40 W m 2 ) for OLR (top, Wm -2 ) Temperature (contours,.1 °C), red positive from Straub and Kiladis 2002

Zonal Wind at Majuro (7  N, 171  E) Regressed against Kelvin-filtered OLR (scaled -40 W m 2 ) for OLR (top, Wm -2 ) Zonal Wind (contours,.25 m s -1 ), red positive from Straub and Kiladis 2002

Specific Humidity at Majuro (7  N, 171  E) Regressed against Kelvin-filtered OLR (scaled -40 W m 2 ) for from Straub and Kiladis 2002 OLR (top, Wm -2 ) Specific Humidity (contours, 1 X g kg -1 ), red positive

Meridional Wind at Majuro (7  N, 171  E) Regressed against MRG-filtered OLR (scaled -40 W m 2 ) for OLR (top, Wm -2 ) Meridional Wind (contours,.25 m s -1 ), red positive

Temperature at Majuro (7  N, 171  E) Regressed against MRG-filtered OLR (scaled -40 W m 2 ) for OLR (top, Wm -2 ) Temperature (contours,.1 °C), red positive

Specific Humidity at Majuro (7  N, 171  E) Regressed against MRG-filtered OLR (scaled -40 W m 2 ) for OLR (top, Wm -2 ) Specific Humidity (contours, 1 X g kg -1 ), red positive

Haertel and Kiladis 2004 Wave Motion

Haertel and Kiladis 2004 Wave Motion

Haertel and Kiladis 2004 Wave Motion

Haertel and Kiladis 2004 Wave Motion

Zonal Wind at Honiara (10  S, 160  E) Regressed against MJO-filtered OLR (scaled -40 W m 2 ) for OLR (top, Wm -2 ) U Wind (contours,.5 m s -1 ), red positive OLR Pressure (hPa) from Kiladis et al. 2005

Temperature at Honiara (10  S,  E) Regressed against MJO-filtered OLR (scaled -40 W m 2 ) for OLR (top, Wm -2 ) Temperature (contours,.1 °C), red positive OLR Pressure (hPa) from Kiladis et al. 2005

Specific Humidity at Truk (7.5  N,  E) Regressed against MJO- filtered OLR (scaled -40 W m 2 ) for OLR (top, Wm -2 ) Specific Humidity (contours, 1 X g kg -1 ), red positive OLR Pressure (hPa) from Kiladis et al. 2005

Q1 Regressed against MJO-filtered OLR over the IFA during COARE from Kiladis et al. 2005

Morphology of a Tropical Mesoscale Convective Complex in the eastern Atlantic during GATE (from Zipser et al. 1981) Storm Motion

Observed Kelvin wave morphology (from Straub and Kiladis 2003) Wave Motion

Two day (WIG) wave cloud morphology (from Takayabu et al. 1996)

from Morita et al., 2006

Equatorial Wave Morphology All waves examined have broadly self-similar vertical structures in terms of their dynamical fields (temperature, wind, pressure, diabatic heating) Cloud morphology is consistent with a progression of shallow to deep convection, followed by stratiform precipitation Suggests a fundamental interaction between wave dynamics and convection across a wide range of scales All waves examined have broadly self-similar vertical structures in terms of their dynamical fields (temperature, wind, pressure, diabatic heating) Cloud morphology is consistent with a progression of shallow to deep convection, followed by stratiform precipitation Suggests a fundamental interaction between wave dynamics and convection across a wide range of scales

Convection in General Circulation Models Question: How well do GCMs do in characterizing intraseasonal tropical convective variability? Jialin Lin et al. (2006) applied identical space-time spectral techniques to observed and modeled tropical precipitation Models used are the 14 coupled ocean-atmosphere GCMs used for intercomparison in the 4th Assessment Report of the Intergovernmental Panel on Climate Change (IPCC) Question: How well do GCMs do in characterizing intraseasonal tropical convective variability? Jialin Lin et al. (2006) applied identical space-time spectral techniques to observed and modeled tropical precipitation Models used are the 14 coupled ocean-atmosphere GCMs used for intercomparison in the 4th Assessment Report of the Intergovernmental Panel on Climate Change (IPCC)

Rainfall Power Spectra, IPCC AR4 Intercomparison 15S-15N, (Symmetric) from Lin et al., 2006 Observations

Rainfall Power Spectra, IPCC AR4 Intercomparison 15S-15N, (Symmetric) from Lin et al., 2006

Rainfall Spectra/Backgr, IPCC AR4 Intercomparison 15S-15N, (Symmetric) from Lin et al., 2006 Observations

from Lin et al., 2006 Rainfall Spectra/Backgr, IPCC AR4 Intercomparison 15S-15N, (Symmetric)

Rainfall Spectra/Backgr., IPCC AR4 Intercomparison 15S-15N, (Antisymm.) from Lin et al., 2006 Observations

Rainfall Spectra/Backgr., IPCC AR4 Intercomparison 15S-15N, (Antisymm.) from Lin et al., 2006

Outstanding Issues General Circulation Models do a relatively poor job in correctly simulating variability in tropical convection (but not necessarily its mean state) Is this due to the misrepresentation of convection itself, or its coupling to the large scale (or both)? Is convection even parameterizable in models? Improvements in the representation of tropical convection will lead to improvements in medium-range weather forecasts in mid-latitudes (and perhaps to ENSO) What is the impact of poor tropical variability in GCMs on climate change scenarios? General Circulation Models do a relatively poor job in correctly simulating variability in tropical convection (but not necessarily its mean state) Is this due to the misrepresentation of convection itself, or its coupling to the large scale (or both)? Is convection even parameterizable in models? Improvements in the representation of tropical convection will lead to improvements in medium-range weather forecasts in mid-latitudes (and perhaps to ENSO) What is the impact of poor tropical variability in GCMs on climate change scenarios?