Transcript Climate response to dust

Climate response to dust
N. Mahowald, M. Yoshioka, D. Muhs,
W. Collins, A. Conley, C. Zender, D.
Fillmore, D. Coleman, P. Rasch
Desert dust/mineral aerosols
• Soil particles suspended in area
• Source:
– Unvegetated, dry soils with strong winds
• Removal
– Dry deposition, especially gravitation settling
– Wet deposition, during precipitation
• Model the sources, transport and deposition processes
in 3-dimensional model (offline transport model:
MATCH/NCEP or NCAR Community Atmospheric Model
(CAM3) from the Community Climate System Model
(CCSM3))
• Papers available at
www.cgd.ucar.edu/tss/staff/mahowald
U. Miami data, Mbouru et al., visibility data
We know that dust responds to climate on regional
to global on short to long time scales by a factor of 4
to 100. Dust is often used a proxy for climate
change in the paleo record.
How does this change in dust impact climate?
Sensitivity study (Yoshioka et al., in
press)
• Using CAM3 and slab ocean model
• AMIP runs (SST impacts)
• Vegetation changes (force model to change similar to
estimated changes 1960s to 1990s)
• Green house gas changes (2x co2 SOM runs)
• Dust changes (with and without dust direct radiative
forcing)
– Only include direct radiative effects (ignoring CCN or IN
interactions, which may be important)
– Can’t get dust signal with amip and vegetation changes—need
to force model to capture dust change at barbados
• Model error
• Land use source of dust
• Vegetation change source of dust
Impacts of dust onto climate/precipitation
CAM3/SOM
Dust radiative
feedback impacts
on precip
Not including
long wave in
CCSM3
enhances double
“ITCZ” (not
shown)
Impacts on Sahel precip.
•SSTs ~50% of observed precip
change
•Vegetation changes Not significant
•GHG has wrong signal (increases
precip in Sahel)
•Model can’t capture dust changes
observed, but observed dust changes
(when forced onto model) cause ~30%
change in observed precip in Sahel
•Dust could be important feedback on
Sahel precip Yoshioka et al., in press
Yoshioka et al., in press
Climate response to dust under
different climates
How robust is this response?
• Physical parameterizations or physical biases will impact
our simulation of dust (or x variable we are interested in).
• How does this impact our precipitation sensitivity?
• Shift in precip due to dust radiative forcing is not
sensitive to climate in our model (Mahowald et al., 2006)
• Response is sensitive to single scattering Albedo (Miller
et al., 2004; other papers)
– Radiative properties of dust are not well established
• Dust absorbs and scatters in long AND short wave
• NOT spherical particles!
Miller et al., 2004
Precip is sensitive to single
scattering albedo
If single scattering albedo is
larger than their base case
(as in our case), see
consistent shifts, maybe.
New version of GISS model
has higher single scattering
albedo than used in Miller et
al., 2004. , due to updates in
optical properties.
How sensitive is the response of ‘desert regions’ to climate
model used?
•Use PCMDI models (17)
•Use anomalies from current climate to force BIOME4 vegetation model (Kaplan
et al., 2003). (Asynchronous coupling, desert vegetation reaches equilibrium
quickly)
•Calculate estimated desert area for future climate
•(use co2 fertilization and no fertilization).
Lots of spread!
CCSM is somewhat
extreme in wetting
Sahel in future
(Regional model
have different
response?)
Want to do this
interactively in model
(Andrea Sealy in
working on
dust/AOVM modeling
now within CCSM)
Latitude
Latitude
Latitude
Mahowald et all, in prep
Latitude
Latitude
Latitude
Smaller scale interactions
• Dust and easterly waves
– 20-40% of dust is generated and transported
associated with easterly waves (Jones et al., 2003;
using NCEP and NCEP/MATCH)
– Easterly waves maybe enhanced by dust (Jones et
al., 2004 NCEP/MATCH; Jones et al in prep (CAM3
T85)
• Dust and hurricanes
– Dust cools surface and suppresses precip in our
model, some observation studies….(Yoshioka et al.,
in press, Wong and Dessler, Evans et al., in prep)…
Summary/conclusions
In this set of model simulations:
• SSTs are responsible for 50% of the Sahel signal (pretty robust
across models)
• Vegetation NS (different models show different results)
• Dust responsible for up to 30% of Sahel drought signal in this model
(consistent with one existing study? Need more models!)
– Dust could be ‘natural’ or anthropogenic (Mahowald et al., 2002;
Prospero and Lamb, 2003; Mahowalld and Luo, 2003; Tegen et al.,
2004; Mahowald et al., 2004)
•
GHG in this model lead to higher precip—not robust result
• Dust potentially an important feedback factor that should be better
explored.
•
•
Not discussed here at length, but should not be ignored:
Anthropogenic changes in ‘natural’ aerosol are potentially large and should not be ignored
–
•
our estimates: (Mahowald et al., 2006; Mahowald and Luo, 2003)
–
–
•
from direct perturbation of land (land use), climate change or carbon dioxide fertilization of plants
PreindustrialI to present (-0.1 to 0.30C)
Present to future (doubled CO2) is about ~+0.06C
Dust changes could also be driving changes in ocean biogeochemistry and carbon dioxide fluxes
(Mahowald et al., 2006; Moore et al., 2006)
Dust response to climate
Percentage change in source area
CCSM3 Dust source (Tg/year)
40.00
20000
20.00
0.00
%
LGM
-20.00
-40.00
PI
DCO2
BASE
BASE-CO2
18000
Constant Veg.
16000
BASE-CO2
14000
BASE
12000
Tune1
10000
Tune2
8000
6000
4000
2000
0
-60.00
LGM
Pre-industrial
Current
Doubled-CO2
-80.00
•Assume only ‘natural’ sources of dust (can’t eliminate 0-50% potential contribution
from land or water use (Mahowald et al., 2002: 2004; Mahowald and Dufresne,
2003; Mahowald and Luo, 2003), but ignore for now).
•Assume climate (precip, Ts, cloudiness) and carbon dioxide fertilization of plants
important. (Smith et al., 2000; Moore et al., in press suggests carbon dioxide
fertilization reasonable with 2x co2).
•Vegetation response most important in this model (similar to Mahowald et al., 1999;
dissimilar to Werner et al.,2002). No dynamic veg (asynchronous coupling with
BIOME3).
Compare model changes to obs
•
•
Can’t distinguish preindustrial from current (Mahowald and Luo, 2003;
Mahowald et al. 2006.
Compare against all available data in current climate. Dust deposition
records for last glacial maximum.
•For LGM: Use geological record to
infer ‘glaciogenic’ sources and best
Current: SOMB
match terrestrial sediment record
SOMBLGMC
SOMBLGMT
Linearity in response in RF (surface or top of atmosphere) or global
surface temperature or global precipitation in this model
Squares TOA, triangles SFC
TOA radiative forcing vs. Surface Temperature response
0.2
0
-2
-1.5
-1
-0.5
0
0.5
1
-0.2
-0.4
Ts (K)
True in other models?
Compare to sensitivity
experiments in the single
scattering albedy done in GISS
model (values courtesy of R.
Miller: Miller and Tegen, 1998;
Miller et al., 2004)
-0.6
-0.8
-1
-1.2
TOA (W/m2)
GISS
CAM