Transcript ppt

North American Hydroclimate Variability in
IPCC’s 20th Century Climate Simulations
Alfredo Ruiz-Barradas
Sumant Nigam
Department of
Atmospheric and Oceanic Science
University of Maryland
2006 Joint Assembly Meeting
Baltimore, Maryland
May 23, 2006
Motivation
To better know the structure and mechanisms of
precipitation variability in state-of-the-art
climate models
At issue:
– Model validation
– Relative contributions of local (evaporation) and
remote (moisture fluxes) water sources
Outline
•
•
•
•
The data sets.
Annual Cycle of Precipitation
Precipitation variability over the Great Plains.
Structure of hydroclimate fields and their
relative contributions associated with
precipitation anomalies.
• The surface temperature signature.
• Conclusions
Data Sets
• US-Mexico retrospective precipitation analysis:1951-1998.
• North American Regional Reanalysis (NARR): 1979-1998.
• Simulations of the Climate of the 20th Century: 1951-1998.
– Historical Forcing:
•
•
•
•
Solar irradiance
Volcanic and anthropogenic aerosols (optical depth)
Ozone, Carbon Dioxide
Other well mixed greenhouse gases
• Resolution:
– Horizontal: R30 (96x80) Gaussian grid
– Vertical: 17-pressure levels
MODELS
Representation/Resolution
Model
Horizontal
Atmosphere
Vertical
Ocean
Atmos.
1x1
Equatorward
of 30 as fine
as 1/3
24
(360x200)
MOM4P0 or
OM3
1x1
Equatorward
of 30 as fine
24
as 1/3
(Sigma?)
(360x200)
MOM4P0 or
OM3
Ocean
Time period
Data Available
Data
at PCMDI
Retrieved
Reference
Web Reference
Jan 1951 to
Dec
Delworth et. al. http://nomad
2000(mrso (submitted
s.gfdl.noaa.go
and mrro look 2004)
v
suspect..)
GFDL (CM2.1)
2.5x2
(144x90
N45L24)
AM2
GFDL (CM2.0)
2.5x2
(B grid?
144x90
N45L24)
AM2
GISS (C4x3)
4x3
(B grid
90x60)
4x3??
12
16
Jan 1850 to
Dec 1999
Jan 1951 to
Dec 2000
GISS (E-H)
5x4
(B Grid
72x45)
2x2 HYCOM
20
16
Jan 1880 to
Dec 1999
Jan 1950 to
Dec 1999
GISS (E-R)
5x4
(B Grid
72x45)
4x5
20
13
Jan 1880 to
Dec 2003
Jan 1950 to
Dec 2003
(gx1v3??)
POP1.4.3
26
Jan 1870-Dec
1999
Jan 1950 to
Dec 1999
Collins, W.D. et
http://www.c
al. ,(2005) J.
csm.ucar.edu
climate
18
Jan 1890-Dec
1999
Jan 1950 to
Dec 1999
Washington,
http://www.c
W.M,et. al.
gd.ucar.edu/p
(2000) Climate
cm
Dynamics
NCAR(CCSM3)
NCAR(PCM)
~1.4x1.41
(Spectral
T85
256x128)
128x64
( Spectral
T42)
CCM3.6.6
384x288
POP1.0
50 (Top
of 220m
has 22
levels)?
Jan 1861 to
Dec 2000
50 (Top
of 220m
has 22
levels)
Jan 1861 to
Dec 2000
Jan 1951 to
Dec 2000
32
Delworth et. al. http://nomad
(submitted
s.gfdl.noaa.go
2004)
v
http://aom.gis
s.nasa.gov/do
c4x3.html
http://www.gi
ss.nasa.gov/t
ools/modelE/m
odelE.html
http://www.gi
ss.nasa.gov/t
ools/modelE/m
odelE.html
MODELS
Model
Representation/Resolution
Vertical
Horizontal
Atmos. Ocean
Ocean
Atmosphere
HADCM3, UK
~3.75x2.5(9
1.25x1.25
6x73)
MICRO3
HIRES
0.28125 X
T106L56~1.
0.1875 47
25x1.125(32
vertical
0x160)
levels
19
56
20
47
Time period
Data
Data Available
Retrieved
at PCMDI
Jan 1860-Dec
1999
Jan 1900-Dec
2000
Jan 1950-Dec
1999 (note
Dec 1999 is
missing for
Ua,Va,ta and
hus). Also, 950
is the level
instead of
925..and 70mb
and 20mb are
missing ie.
there are 15
vertical levels
only)
Jan 1950-Dec
1999
Reference
Gordon, et al
(2000) (16,
147-168);
Johns, C.T,
(1997) Clim.
Dyn (13, 103134).
Web Reference
http:/www.me
toffice.com/re
search/hadley
centre/models
/HadCM3.html
http://www.c
csr.utokyo.ac.jp/ky
osei/hasumi/M
IROC/techrepo.pdf
Precipitation:
Annual Cycle &
Annual Mean
(1979-1998)
Maximum over
northwestern US
In January
Models do well
over US northwest
in winter
But seem to have some
problems over
southeastern and central
US in spring and summer
The annual cycle diminishes and
occurs earlier in the summer months
from tropical Mexico to central US
Weak maximum over
southern US in winter/spring
months
JJA Precipitation
Climatology
Precipitation: JJA STD
Great Plains
Precipitation Index is
defined over the
area of maximum
STD in observations
1.00 mm/day
US-MEX
0.91 mm/day
What makes the
observed and
Simulated STD?
0.63 mm/day
min
1.12 mm/day
0.92 mm/day
Max
0.72 mm/day
0.78 mm/day
Histogram of
Precipitation Events
According to GPP
Indices (1951-1998)
0.91 mm/day
76<0
68>0
0.63 mm/day
1.00 mm/day
77<0
Large STD
due to extreme
events
67>0
67<0
1.12 mm/day
75<0
69>0
0.92 mm/day
72<0
0.72 mm/day
73<0
71>0
Small STD due to
77>0 a concentration of
small events
72>0
0.78 mm/day
73<0
71>0
Regressed Precipitation
Anomalies on JJA
GPP Indices
1.04 mm/day
1.13 mm/day
0.73 mm/day
0.89 mm/day
0.63 mm/day
0.92 mm/day
0.78 mm/day
Regressed Convective
Precipitation on
GPP Indices
Regressed Moisture Flux
Anomalies on JJA
GPP Indices
0.66 mm/day
0.45 mm/day
0.46 mm/day
0.40 mm/day
0.83 mm/day
0.91 mm/day
0.66 mm/day
Regressed Evaporation
Anomalies on JJA
GPP Indices
0.22 mm/day
0.69 mm/day
0.04 mm/day
0.59 mm/day
0.14 mm/day
-0.04 mm/day
0.06 mm/day
CI=1/3 of that in P & MFC
So far, we have the following picture regarding the
relative controls of precipitation variability:
•Evaporation dominates over Moisture Flux Convergence:
CCSM3, GFDL-CM2.1
•Moisture Flux Convergence largely dominates over Evaporation:
GISS-EH, MIROC3.2(hires)
•Moisture Flux Convergence dominates over Evaporation:
NARR/US-MEX
PCM,UKMO-HadCM3
Is precipitation recycling the same in CCSM3 and GFDL-CM2.1?
Autocorrelation of GPP Indices
Standard Errors
Significance at the
0.05 level
CCSM3 & GFDL-CM2.1
do not recycle
precipitation in
the same way!!
Regressed Surface
Temperature Anomalies
on JJA GPP Indices
Large evaporation anomalies
-0.8 K have implications on the
surface energy balance
and so on surface
temperature, the variable of
choice when analyzing warming
scenarios
-2.4 K
-0.3 K
-1.9 K
-0.9 K
-0.0 K
-0.7 K
Conclusions
•NARR/US-Mexico data sets suggest that remote water sources
(moisture fluxes) dominate over local water sources (evaporation) in
the generation of interannual rainfall variability over the Great Plains
during the warm-season.
Three different hierarchy of process in models:
 MFC >> E: MIROC3.2(hires), GISSEH, GISS-ER
 MFC > E: UKMO-HadCM3, PCM
 E > MFC: CCSM3, GFDL-CM2.1, GFDL-CM2.0, GISS-AOM
•Deficient simulation of moisture pathways feeding the Great Plains.
•In consequence: regional hydroclimate simulations and
predictions remain challenging for global models, at least
in the context of variability over the Great Plains.