Transcript Maslowski`s ASM presentation

Regional Arctic Climate System Model
(RACM) – Project Overview
A 4-year (2007-2010) DOE / SciDAC-CCPP project
Participants:
Wieslaw Maslowski
(PI)
John Cassano
(co-PI)
William Gutowski (co-PI)
Dennis Lettenmeier (co-PI)
Greg Newby, Andrew Roberts,
Juanxiang He, Anton Kulchitsky
- Naval Postgraduate School
- University of Colorado
- Iowa State University
- University of Washington
- Arctic Region Supercomputing
Center
Dave Bromwich (OSU), Gabriele Jost (HPCMO), Tony Craig (NCAR),
Jaromir Jakacki (IOPAN), Mark Seefeldt (CU), Chenmei Zhu (UW),
Justin Glisan (ISU), Jaclyn Kinney (NPS)
IARC / Arctic System Model Workshop, Boulder, CO, May 19-21, 2008
Main science objective
To synthesize understanding of past and present states
and thus improve decadal to centennial prediction of
future Arctic climate and its influence on global climate.
Specific Goals
• develop a state-of-the-art Regional Arctic Climate system
Model (RACM) including high-resolution atmosphere,
ocean, sea ice, and land hydrology components
• perform multi-decadal numerical experiments using high
performance computers to understand feedbacks, minimize
uncertainties, and fundamentally improve predictions of
climate change in the pan-Arctic region
• provide guidance to field observations and to GCMs on
required improvements of future climate change
simulations in the Arctic
Regional Arctic climate model
components and resolution
•
•
•
•
•
Atmosphere - Polar WRF
(gridcell ≤50km)
Land Hydrology – VIC
(gridcell ≤50km)
Sea Ice – CICE/CSIM
(gridcell ≤10km)
Ocean - POP
(gridcell ≤10km)
Flux Coupler – CCSM/CPL7
RACM domain and elevations
(red box represents the domain of ocean and sea ice models)
Pan-Arctic region to include:
- all sea ice covered ocean in the northern hemisphere
- Arctic river drainage
- critical inter-ocean exchange and transport
- large-scale atmospheric weather patterns (AO, NAO, PDO)
Why develop a regional Arctic
climate model?
1. Facilitate focused regional studies of the Arctic
2. Resolve critical details of land elevation, coastline
and ocean bottom bathymetry
3. Improve representation of local physical processes
and feedbacks (e.g. forcing and deformation of sea
ice)
4. Minimize uncertainties and improve predictions of
climate change in the pan-Arctic region
Comparison of sea ice conditions in September 2002
Arctic Sea Ice cover in September 2002
3
3
3
2
2
2
1
1
1
CCSM3(b) simulates too much ice on the
Greenland shelf (1), too much/little melt
in the eastern(2) / western(3) Arctic.
NCAR/CCSM3 case (b30.040b) prediction
of summer ice-free Arctic by 2050
Comparison of areal sea ice fluxes through Fram Strait
-CCSM3 sea ice export about twice as high as compared to
Kwok et al. (2003) and NPS/NAME fluxes
- Possibly too strong atmospheric forcing at Fram Strait
- Consequences include
- too much ice production in the Arctic Ocean
- overestimate of buoyancy flux into the North Atlantic
25-year mean ocean volume transport (Sv) / heat transport (TW)
Note: 1Sv = 10 m6/sec; 1TW = 3.6 Petajoules/hour or 86.4 Petajoules/day or 2592 Petajoules/month
CCSM3(b)
NPS/NAME
In
Out
Net
In
Out
Net
Fram
Strait
2.0/17
-6.9/ -23
-4.9/ -6
6.0/45
-8.4/ -36
-2.4/ +9
Barents
Sea
Opening
4.8/115
-0.3/ -5
4.5/110
5.0/107
-1.8/ -28
3.2/79
FJL-NZ
4.7/32
-0.35/ -1
4.35/31
3.4/2.9
-0.8/ -0.7
2.6/2.2
NPS/NAME TRANSPORTS (Maslowski et al., JGR, 2004)
Fram Strait ‘in’ obs estimates: 7.0 Sv / 50 TW - Courtesy of A. Beszczynska-Möller, AWI
FJL-NZ: near-zero heat transport (Gammelsrod et al., JMS submitted)
OCEAN BATHYMETRY/RESOLUTION IMPACTS
18-km Model
0-225 m (levels 1-7), every vector
9-km Model
0-223 m (levels 1-15), every 2nd vector
• Barents Sea outflows (north of Novaya Zemlya and through Kara Gate) look similar but:
• Mean paths significantly different due to representation of bathymetry (I.e. resolution)
• Velocity magnitudes differences
• 9-km model circulation shown to match observed well (Maslowski et al., 2004)
• Implications for location of fronts, water mass transformations, heat and salt balances
(from Maslowski et al., 2008)
LAND TOPOGRAPHY / RESOLUTION IMPACTS
ERA40 Annual Precipitation
ERA40 Annual Precipitation
• Increased horizontal resolution
allows for improved representation
of topography
• Topography impacts atmospheric
circulation, precipitation,
temperature, etc.
Polar MM5 Annual Precipitation
• ERA40 precipitation (above) is
“smoothed” compared to higher
resolution (50 km) Polar MM5
simulation (right)
• This will impact both atmosphere
and land/ocean
Cyclone Central Pressure and Size
• Model resolution impacts the size and intensity of cyclones
• Comparison of AMPS ( 20 km; based on Polar MM5 and
WRF) and three coarser reanalyses in the Southern Ocean
• AMPS simulates lower pressure in and smaller cyclones
than all reanalyses
• Similar results are expected in Arctic
Finer resolution captures dispersed features
missed by coarse grids
Wetlands
RMS Diff. [m]
RMS Difference
vs. RMSD
Baseline
(500
hPa Heights)
500 hPa
vs.
Standard
100
WET10
Wetlands
cases
WET30
80
LBCL
Unforced
LBCS
ICL
“noise”
ICS
60
40
20
0
0
10
20
30
Day
Gutowski et al. (2007)
40
50
60
Sea Ice Divergence
near SHEBA Tower
(Stern & Moritz, 2002)
Ice strain (reds/yellows)
(200 km x 200 km window)
Attribution of observed trends in Eurasian Arctic river runoff:
Why don’t model reconstructed trends match observations?
• Long-term
streamflow changes
(Peterson et al.,
2002) are not
captured by model
in permafrost basins
(particularly in
discontinuous
permafrost).
• Reasons include
improper
permanent ground
ice initialization and
lack of tracking.
• Improvements to the VIC frozen soils algorithm to handle
permafrost are underway.
(Visuals courtesy of Jennifer Adam, Washington State University)
High Quality GHCN
Precipitation Stations
High Quality GHCN
Temperature Stations
RACM 2008-2009 Outlook
1. Evaluate uncoupled model simulations for
physical and numerical optimizations in
RACM
2. Couple each climate model component to the
coupler (CPL7)
3. Run and validate results as in #2
4. Couple all climate model components and
run tests with RACM