The Climatology and Interannual Variability of North American Stormtracks in the GFS and CFS Global Climate Models

1
The Climatology and Interannual Variability of
North American Stormtracks in the GFS and CFS
Global Climate Models
Timothy Eichler and Wayne Higgins Climate
Prediction Center/NCEP
2
GFS Simulation

• T62 GFS run with AMIP II SST
• Run from 1950-2002
• SLP saved twice daily

CFS Simulation

• 2.5x2.5 fully coupled simulation
• Run from 2002-2033
• Free Run (i.e. climate mode)
• SLP saved twice daily

3
Storm Track Frequency Climatology (5x5 grid) for
I Obs II GFS III CFS from a Winter through
d Fall
I
III
II
4
Storm Track Frequency Difference a GFS-OBS b
CFS-OBS
a
b
5
Storm Track Frequency Difference CFS-GFS
6
SLP (hPa) of Storms I Obs II GFS III CFS
a-d (winter through fall)
I
II
III
7
SLP Diff I GFS-Obs II CFS-OBS (a-d
winter through fall)
I
II
8
Standard Deviation Analysis for I GFS and II
CFS
a
I
II
9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
Composite Stormtrack Frequency Anomaly for a
Strong El Nino-neutral and b strong La
Nina-neutral
a
b
13
Composite GFS Stormtrack Frequency Anomaly for a
Strong El Nino-Neutral and b Strong La
Nina-Neutral
a
b
14
Composite CFS Stormtrack Frequency Anomaly for a
Strong El Nino-Neutral and b Strong La
Nina-Neutral
a
b
15
Stormtrack difference (Strong El Nino Strong La
Nina) for a Observations b GFS Model c CFS
Model
a
b
c
16
Merdional Temp. Gradient ((degrees C/ km)100)
for a obs b GFS model c GFS-obs
a
b
c
17
Merdional Temp. Gradient ((degrees C/ km)100)
for a obs b CFS model c CFS-obs
a
b
c
18
H500 Gradient ((m/km)100) for a obs b CFS
model c CFS-obs
a
b
c
19
H500 Gradient ((m/km)100) for a obs b GFS
model c GFS-obs
a
b
c
20
Conclusions

• GFS Produces Stormtrack Climatology
  spatially reasonable though approximately 50
  less frequent CFS slightly better.
• Storm tracks less frequent and weaker
  especially in the North Pacific and North
  Atlantic in areas normally associated with strong
  baroclinicity.
• GFS and CFS models exhibit a seasonal
  cycle for stormtracks though weaker than obs.
• GFS and CFS stormtracks show a response to
  ENSO evident especially when comparing strong
  events
• Reduced storm frequency relative to
  observed implies weak model variability.
  Evidence suggests that the GFS and CFS models
  have less baroclincity than observed.

21
Future Work

• Explore GFS and CFS model physics to
  ascertain why models storm variability is less
  than observed (e.g. look at surface heat budget)
  
• Investigate Storm structure (e.g. fronts,
  precipitation, etc.) to see if they are realistic
  
• Use stormtracks software on other data rr
  data, ETA model Program has potential as a
  prognostic tool