storm track response to ocean fronts in a global high-resolution climate model
DESCRIPTION
Department of Energy. Storm track response to Ocean Fronts in a global high-resolution climate model. R. Justin Small, Frank Bryan and Bob Tomas NCAR Young-Oh Kwon WHOI + 2 anonymous reviewers. Aims. Investigate influence of ocean fronts on atmospheric storm track in winter - PowerPoint PPT PresentationTRANSCRIPT
Storm track response to Ocean Fronts in a global high-resolution
climate model
R. Justin Small, Frank Bryan and Bob TomasNCAR
Young-Oh KwonWHOI
+ 2 anonymous reviewers
Department of Energy
Aims• Investigate influence of ocean fronts on
atmospheric storm track in winter– Surface storm track– Free-troposphere storm track
• What are the key storm track statistics that are affected?
• What affects baroclinicity?• Using a global atmospheric climate model
– 1. North Atlantic– 2. Southern Ocean– 3. North Pacific
Experiments• Community Atmosphere Model version 4
– Developed at NCAR, Department of Energy, US labs– Hydrostatic, sigma-coordinate global model– ½ deg. grid spacing, 27 levels (<6 in lowest 1000m).
• Twin experiments, atmosphere-only. – 1. Control has realistic SST in region (e.g. N. Atlantic)– 2. Smooth Global SST experiment– SST is a climatology based on satellite/in situ data (1/4deg.,
Reynolds et al 2007).
• Each run for 60 years to gain some statistical significance.
Methods and Data• We use a high pass filter
– V’=V-<V>– where <V> is 5-day mean at surface, seasonal
mean or monthly mean in free troposphere
• Compute climatological mean of quantities such as <V’V’> , <V’T’>
• Apply smoothing to SST fields as boundary condition for AGCM.
• 1• 4000 passes of 1-4-1 filter• 1
• Comparisons are made with ERA-INTERIM data 1979-2009 (ERA-I) and OAFLUX, QSCAT
North Atlantic case, Boreal winter (DJF).
(a) (b)
(c)
SST FOR CONTROL
SMOOTH SST EXPERIMENT
SST differenceSST DIFF
C /100km
(d)
SST gradient difference
Frequency distribution of strong SST gradients
High-res coupled model
Low-res coupled modelReynolds OI
SST 1/4deg.
Heavy smoothing of Reynolds OI SST 1/4deg.
Light smoothing of Reynolds OI SST 1/4deg.
Histograms of occurrence of binned SST gradients within 1deg. C/100km contour in North Atlantic including Gulf Stream. Uses data from DJF climatology. Units deg.C per 100km.
What storm track quantities are significantly affected by ocean
front and what quantities are not?
10-6s-1
10-6s-1
(a) (b)
(c)
control
SMTH
10-6s-1
(d)
Standard deviation of near –surface transient eddy vorticity variability. Filtered to retain only timescales less than 5 days. Note that differences (control-smooth, bottom right panel) of std. dev (’) overly SST anomalies, and reach up to 30% of smooth value.
10-5s-1
Std.dev(’) Control
Std.dev(’) Smooth
Diff in Std.dev(’) +SST anomaly
30%
OAFLUX obs- Joyce and Kwon 2009
Relative vorticity variability
SEA LEVEL PRESSURE VARIABILITTY. SLP sub-5day variability and differences
Add significance
SMOOTHCONTROL
DIFFERENCE
hPa hPa
hPahPa
Surface geostrophic vorticity sub-5day variability and differences
2
2
2
21
11
y
p
x
p
f
y
p
fyx
p
fxuy
vx ggg
GEOSTROPHIC VORTICITY variability.DERIVED JUST FROM SLP.
SMOOTHCONTROL
DIFFERENCE
10-6s-110-6s-1
10-6s-1
25-30%
Atlantic DJF: Meridional Heat Flux
V’T’ Control V’T’ ERA-I
ms-1K ms-1K
30%
In the right panel only differences significant at 95% are shown, and contours show SST differences of +/- 2 C from Fig. 1c. The number shown is the approx. ratio of the amplitude of the difference to the amplitude of the maximum in the smooth case, expressed as a percentage.
V’T’ ERA-I
ms-1K
V’T’ Diffn.Control-Smooth
ms-1Cms-1K
V’T’
ms-1K
V’T’ Control
25%
V’T’ ERA-I
ms-1K
Control-Smooth
Transient eddy meridional heat flux 500hPa
Transient eddy meridional heat flux 850hPa
m2s-2
Atlantic DJF: Meridional wind variance
V’V’ V’V’ Control
m2s-2 m2s-2
10%
V’V’ ERA-I
m2s-2
Low-light – CAM wind variance (& heat-flux) is too high compared to ERA-I and MERRA . Therefore adding the ocean front worsens the comparison.
Control-Smooth
V’V’ Control V’V’ ERA-I
m2s-2 m2s-2
15%V’V’ ERA-I
m2s-2
V’V’
Transient eddy meridional wind variance 850hpa
7% std. dev
Control-Smooth
Transient eddy meridional wind variance 500hpa
BAROCLINICITY
– WHAT COUNTERS THE EFFECT OF EDDIES IN REMOVING TEMPERATURE GRADIENT?
– Latent heat release over western boundary currents helps maintain baroclinicity (Hoskins Valdes 1990,JAS)
– Sensible heating maintains baroclinicity and anchors storm track (Nakamura et al 2008 GRL, Nonaka et al 2009, Sampe and Nakamura 2010 JCLIM, Ogawa et al 2012, Hotta and Nakamura 2011)
Baroclinicity
Eady (1949)- growth rate of most unstable mode
Baroclinicity
0
31.0N
g
Differences reduce to 7% at 500hPa
(a) (b) (c)
(e) (f)(d)
50%
30%
950hPa
850hPa
SMTH
SMTH
ATL
ATL
Thermodynamic potential temperature budget at 950hPa. Units degC./day.DJF climatology mean (from 10 years)
HOR. ADVECTION VER. ADVECTION
-d/dy V’T’ -d/dz W’T’
Condensational Heating sensible Heating
BOUNDARY LAYER HEAT BUDGET –
Heat budget at 850hPaHOR. ADVECTION VER. ADVECTION
-d/dy V’T’ -d/dz W’T’
Condensational Heating sensible Heating
Thermodynamic potential temperature budget at 850hPa. Units degC./day.DJF climatology mean (from 10 years)
FREE TROPOSPHERE HEAT BUDGET –
Vertically integrated total eddy heat flux divergence (color), for a) the SMTH case, b), ATL case and c) their difference. The corresponding climatological SST is shown as contours in a, b) and SST difference in c).
SEE KWON AND JOYCE PRESENTATION
(a) (b)
(c)
ATLSMTH
TRANSIENT EDDY HEAT FLUX DIVERGENCE – CONTROL BY OCEAN FRONT
A few results from Southern Ocean focusing on South Indian Ocean.
ms-1K
25%/83%
SOUTHERN OCEAN CASE, JJA. Relationship of transient eddy heat flux to SST gradient.
(a)V’T’ 850 DIFF
(f)
33%BAROCLINICITY DIFF
(b)
(SMOOTH)
C /100km
(c)
SST DIFFERENCE
SST GRAD DIFF
(a)
(b)
SOUTHERN OCEAN CASE. Effective eddy diffusivity- eddy heat flux divided by mean temperature gradient
10-5m2s-1
10-5m2s-1
SMTH CASE
CONTROL CASE
Mean circulation response and interannual variability
20% reduction of zonal wind
Fig. 1. Circulation response in the North Atlantic in DJF. a, c, d) show diffeernce between the ATL and SMTH runs for a) The sea level pressure, c) the 950hPa zonal wind and d) 500hPa geopotential height. Here stipling denotes 95% significance according to the t-test. b) shows the climatological mean zonal wind at 950hPa in the SMTH case.
hPa
gpmms-1gpm
(a) (b)
(c) (d)
ms-1
SEA LEVL PRESSUREDIFF
U950 MEAN
U950DIFF
Z500 DIFF
30%
Fig. 2. a) The climatological mean 250hpa zonal wind (U250) in the SMTH case for DJF over the North Atlantic. B) the difference in standard deviation of U250 between ATL and SMTH run. Stipling in b) denotes 95% significance according to the f-test.
(a)
(b)
ms-1
ms-1
U250 MEAN
DIFF IN U250 INTERANNUAL STANDARD DEVIATION
Conclusions• Ocean fronts induce large (~30%) changes in heat
flux, moisture flux (~40%), and precip. in winter– Reaching well above the boundary layer – to > 500hPa– vorticity variance at surface (~30%)
• Smaller influence on wind (~10%) and sea level pressure (few% locally) variance
• Baroclinicity and eddy heat flux• Maintained by sensible heating in boundary layer• Condensational heating above that• In Southern ocean v’T’~ dT/dy
• Results may be (very) model-dependent
Heat budget at 950hPa – diff unsmoothed minus smoothed
Heat budget at 850hPa – diff unsmoothed minus smoothed
HOR. ADVECTION VER. ADVECTION
-d/dy V’T’ -d/dz W’T’
Condensational Heatingsensible Heating
Sens. Ht.
Fig. 14. Surface heating and tropospheric temperature differences between ATL and SMTH run. a, d, g) show surface sensible heating, surface latent heating, and precipitation respectively. b), e, and h) show temperature tendency at 950hPa (from sensible heat), and 850hPa and 500hPa (from condensational heating.) c), f), and I) show the corresponding air potential temperature. In right panels the corresponding SST anomalies of +2C(-2C) are shown as thick (thin) solid lines.
(a) (b)
T850 diff.
(e)
(b) (c)
(d) (f)
(g) (h) (i)
Total Prec.
dT/dt 950hPa
dT/dt 850hPa
T 950hPa
T 850hPa
dT/dt 500hPa T 500hPa
Lat. Ht.
Wm-2
Wm-2
mmday-1
Kday-1
Kday-1
Kday-1
Methods
ERA, 5 dayERA, 30 day ERA, 90 day
ERA-INTERIM “heat flux” V’T’ for DJF for different frequency bands
ms-1Kms-1Kms-1K
Frequency response
25deg.
Discussion
• Results get slightly shaky from here on…
To+2
To-2
T’=4
(a)
Figure 17. Schematic showing scenarios for increases to v’T’ due to an ocean front. The solid lines are hypothetical isotherms deliniating a kink in a baroclinic zone (developing into a cold and warm front.) In a), b) there is no notable change to the displacement of the isotherm (no change to v’)
Strong baroclinicity
To+
To-
T’=2
(b)
Weak baroclinicity
Noting that v’T’ ~ baroclinicity (T) leads to possible:Mechanism 1. Mixing length.
t=t0 t=t1 t=t2
t=t0 t=t1 t=t2
(c)
(d)
Figure 17. Schematic showing possible scenarios for increases to v’T’ due to an ocean front. The solid lines are hypothetical isotherms deliniating a kink in a baroclinic zone (developing into a cold and warm front.) In c), d) there is a notable change to the displacement of the isotherm (change to v’). In c, d) a stronger baroclinicity leads to larger growth rate and displacements leads to larger changes in v’T’, particularly later in wave development.
Weak baroclinicity
Strong baroclinicity
Eady growth rate would suggest…Mechanism 2. growth rate ~ T
A note on dynamical fields
Sea Level Pressure sub-5day variability and differences
Add significance
hPa hPa
hPa hPa
ATL ATL SMTH
Surface geostrophic V sub-5day differences
ms-1
11%
Surface geostrophic vorticity sub-5day differences
10-5s-1
30%
ALL GEOSTROPHIC
Comparison with Aquaplanet
Nakumura, Sampe et al. 2008, Sampe et al. 2010.
Maximum SST gradient changed by factor ~6 in zonal mean.SST anomalies 5C or more (all one-sign) in zonal mean.
Maximum SST gradient changed by factor ~3 locally (smaller in zonal mean).SST anomalies up to 5C locally, more typically 2C or less and have both signs.
C per 100km
C /100kmC /100km
SMOOTH CONTROL
Conclusions• Ocean fronts induce large (~30%) changes in heat flux, moisture
flux (~40%), vorticity variance, and precipitation in winter– Reaching well above the boundary layer – to > 500hPa
• Smaller influence on wind (~10%) and sea level pressure (few% locally) variance
• Comparison with reanalysis:– Model Heat flux agrees well with ERA-I and MERRA in Atlantic, is too
high in Southern Ocean– Model wind variance is too high in both regions
• Results may be (very) model-dependent
Way ahead
Moving to CAM-5, high-resolution– Improved convection schemes etc.– ¼ deg, 30 levels– maybe 1/8deg, more vertical levelsCoupled simulations– ocean model 1/10th deg. Parallel Ocean Program (POP)– 40+ years so farSpatial filtering on-line in model coupler (for SST, fluxes etc.) Show animation (if audience still awake)
Do ocean fronts influence storm tracks? • Strong influence
– Latent heat release over western boundary currents helps maintain baroclinicity (Hoskins Valdes 1990)
– Ocean fronts essential to eddy variability associated with polar front jet (Nakamura et al 2008)
• Moderate Influence• Ocean dynamics shifts location of storm track (Wilson et
al. 2009, Brayshaw et al. 2011)
• No influence– Self-maintenance, eddies and mean jet, no role of ocean
(Robinson 2006)
CAM model
Courtesy Joe Tribbia, NCAR
From Minobe et al 2008. Low level convergence proportional to Laplacian of sea level pressure and to Laplacian of SST.
CCSM. From Bryan et al 2010. FIG. 4. Laplacian of sea level pressure (color, 1029 Pa m22) and horizontal convergence of lowest model level wind(contours, interval 2 3 1026 s21, negative values dashed) for the winter season (Nov-Feb) in the Gulf Stream region: high-res CCSM4.
Gulf Stream and atmospheric convection
Vertically integrated total eddy heat flux divergence
Meridional component only i.e. d/dy v’T’ etc
Atlantic DJF: 850hPa
V’T’ Control V’T’ ERA-I
V’V’ Control V’V’ ERA-I
ms-1K
m2s-2
ms-1K
m2s-2
30%
15%
In the right panel only differences significant at 95% are shown, and contours show SST differences of +/- 2 C from Fig. 1c. The number shown is the approx. ratio of the amplitude of the difference to the amplitude of the maximum in the smooth case, expressed as a percentage.
V’T’ ERA-I
ms-1K
V’V’ ERA-I
m2s-2
V’T’ Diffn.
V’V’
Transient eddy “heat flux”
Transient eddy meridional wind variance
7% std. dev
Control-Smooth
Control-Smooth
ms-1C
m2s-2
Atlantic DJF: 500hPa
ms-1K
V’T’
ms-1K
V’T’ Control
V’V’ V’V’ Control
m2s-2 m2s-2
25%
10%
V’V’ ERA-I
m2s-2
V’T’ ERA-I
ms-1K
Transient eddy meridional wind variance
Transient eddy “heat flux”
Low-light – CAM wind variance (& heat-flux) is too high compared to ERA-I and MERRA . Therefore adding the ocean front worsens the comparison.
Control-Smooth
Control-Smooth
Indian Ocean JJA: 850hPa
V’T’ Control
ms-1K ms-1K
30%
V’T’ ERA-I
ms-1K
V’T’ Diffn.
14%
V’V’ Control V’V’ Diff’n
m2s-2 m2s-2
V’V’ ERA-I
m2s-2
In the right panel only differences significant at 95% are shown, and contours show SST differences of +/- 2 C from Fig. 1c. The number shown is the approx. ratio of the amplitude of the difference to the amplitude of the maximum
in the smooth case, expressed as a percentage. Low-light – CAM wind variance (& heat-flux) is too high compared to ERA-I and MERRA . Therefore adding the ocean front worsens the comparison.
ms-1Kms-1K
Transient eddy “heat flux”
Transient eddy meridional wind variance
Indian Ocean JJA: 500hPa
In the right panel only differences significant at 95% are shown, and contours show SST differences of +/- 2 C from Fig. 1c. The number shown is the approx. ratio of the amplitude of the difference to the amplitude of the maximum in the smooth case, expressed as a percentage.
V’T’ Control V’T’ ERA-I
ms-1K ms-1K
25%
V’T’ ERA-I
ms-1K
V’V’ Control V’V’ Diff’nV’V’ ERA-I
15%
m2s-2 m2s-2m2s-2
Transient eddy meridional wind variance
Transient eddy “heat flux”