Happily some of these tests used only sea salt as the models which could also account for the persistent nature of some of the trails. But sadly
others used black carbon
A comprehensive description and evaluation
of the GFDL aerosol simulation are given in
Ginoux et al. (2006). Below are the general
characteristics:
Aerosol fields: The aerosols used in the GFDL
climate experiments are obtained from simulations performed with the MOZART 2 model
(Model for Ozone and Related chemical Tracers) (Horowitz et al., 2003; Horozwitz, 2006).
The exceptions were dust, which was generated
with a separate simulation of MOZART 2, using sources from Ginoux et al. (2001) and wind
fields from NCEP/NCAR reanalysis data; and
sea salt, whose monthly mean concentrations
were obtained from a previous study by Haywood et al. (1999). It includes most of the same
aerosol species as in the GISS model (although
it does not include nitrates), and, as in the GISS
model, relates the dry aerosol to wet aerosol
optical depth via the model’s relative humidity
for sulfate (but not for organic carbon); for sea
salt, a constant relative humidity of 80% was
used. Although the parameterizations come
from different sources, both models maintain a
Sea salt: The GISS model has a much larger sea
salt contribution than does GFDL (or indeed
other models).
Global and regional distributions: Overall, the
global averaged AOD is 0.15 from the GISS
model and 0.17 from GFDL. However, as shown
in Figure 3.8, the contribution to this AOD from
different aerosol components shows greater
disparity. For example, over the Southern Ocean
where the primary influence is due to sea salt in
the GISS model, but in the GFDL it is sulfate.
The lack of satellite observations of the component contributions and the limited available
in situ measurements make the model improvements at aerosol composition level difficult.
Climate simulations: With such large differences in aerosol composition and distribution
between the GISS and GFDL models, one
might expect that the model simulated surface
temperature might be quite different. Indeed,
the GFDL model was able to reproduce the
observed temperature change during the 20th
century without the use of an indirect aerosol
effect, whereas the GISS model required a
substantial indirect aerosol contribution (more
than half of the total aerosol forcing; Hansen
et al., 2007). It is likely that the reason for this
difference was the excessive direct effect in
the GFDL model caused by its overestimation
of the sulfate optical depth. The GISS model
direct aerosol effect (see Section 3.6) is close to
that derived from observations (Chapter 2); this
suggests that for models with climate sensitivity close to 0.75°C/(W m-2
) (as in the GISS and
GFDL models), an indirect effect is needed.very large growth in sulfate particle size when
the relative humidity exceeds 90%.
Global distributions: Overall, the GFDL global
mean aerosol mass loading is within 30% of
that of other studies (Chin et al., 2002; Tie et
al., 2005; Reddy et al., 2005a), except for sea
salt, which is 2 to 5 times smaller. However,
the sulfate AOD (0.1) is 2.5 times that of other
studies, whereas the organic carbon value is
considerably smaller (on the order of 1/2).
Both of these differences are influenced by
the relationship with relative humidity. In the
GFDL model, sulfate is allowed to grow up to
100% relative humidity, but organic carbon
does not increase in size as relative humidity
increases. Comparison of AOD with AVHRR
and MODIS data for the time period 1996-2000
shows that the global mean value over the ocean
(0.15) agrees with AVHRR data (0.14) but there
are significant differences regionally, with the
model overestimating the value in the northern
mid latitude oceans and underestimating it in
the southern ocean. Comparison with MODIS
also shows good agreement globally (0.15), but
in this case indicates large disagreements over
land, with the model producing excessive AOD
over industrialized countries and underestimating the effect over biomass burning regions.
Overall, the global averaged AOD at 550 nm is
0.17, which is higher than the maximum values
in the AeroCom-A experiments (Table 3.2) and
exceeds the observed value too (Ae and S* in
Figure 3.1).
Composition: Comparison of GFDL modeled
species with in situ data over North America,
Europe, and over oceans has revealed that the
sulfate is overestimated in spring and summer and underestimated in winter in many
regions, including Europe and North America.
Organic and black carbon aerosols are also
overestimated in polluted regions by a factor
of two, whereas organic carbon aerosols are
elsewhere underestimated by factors of 2 to 3.
Dust concentrations at the surface agree with
observations to within a factor of 2 in most
places where significant dust exists, although
over the southwest U.S. it is a factor of 10 too
large. Surface concentrations of sea salt are
underestimated by more than a factor of 2.
Over the oceans, the excessive sulfate AOD
compensates for the low sea salt values except
in the southern oceans.
Size and single-scattering albedo: No specific comparison was given for particle size
or single-scattering albedo, but the excessive
sulfate would likely produce too high a value
of reflectivity relative to absorption except in
some polluted regions where black carbon (an
absorbing aerosol) is also overestimated.
As in the case of the GISS model, there are several concerns with the GFDL model. The good
global-average agreement masks an excessive
aerosol loading over the Northern Hemisphere
(in particular, over the northeast U.S. and Europe) and an underestimate over biomass burning regions and the southern oceans. Several
model improvements are needed, including
better parameterization of hygroscopic growth
at high relative humidity for sulfate and organic
carbon; better sea salt simulations; correcting
an error in extinction coefficients; and improved biomass burning emissions inventory
(Ginoux et al., 2006).
3.5.1C. COMPARISONS BETWEEN GISS AND
GFDL MODEL
Both GISS and GFDL models were used in the
IPCC AR4 climate simulations for climate sensitivity that included aerosol forcing. It would
be constructive, therefore, to compare the similarities and differences of aerosols in these two
models and to understand what their impacts are
in climate change simulations. Figure 3.8 shows
the percentage AOD from different aerosol
components in the two models.
Sulfate: The sulfate AOD from the GISS model
is within the range of that from all other models
(Table 3.3), but that from the GFDL model exceeds the maximum value by a factor of 2.5. An
assessment in SAP 3.2 (CCSP 2008; Shindell et
al., 2008b) also concludes that GFDL had excessive sulfate AOD compared with other models.
The sulfate AOD from GFDL is nearly a factor of
4 large than that from GISS, although the sulfate
burden differs only by about 50% between the
two models. Clearly, this implies a large difference in sulfate MEE between the two models.
BC and POM: Compared to observations, the
GISS model appears to overestimate the influence of BC and POM in the biomass burning
regions and underestimate it elsewhere, whereas
the GFDL model is somewhat the reverse: it
overestimates it in polluted regions, and un-from sulfate and POM. This points out the
importance of improving the model ability to
simulate each individual aerosol components
more accurately, especially black carbon.
Separately, it is estimated from recent model
studies that anthropogenic sulfate, POM, and
BC forcings at TOA are -0.4, -0.18, +0.35 W
m-2, respectively. The anthropogenic nitrate
and dust forcings are estimated at -0.1 W m-2
for each, with uncertainties exceeds 100%
(IPCC AR4, 2007).
• In contrast to long-lived greenhouse gases,
anthropogenic aerosol RF exhibits significant
regional and seasonal variations. The forcing
magnitude is the largest over the industrial
and biomass burning source regions, where
the magnitude of the negative aerosol forcing
can be of the same magnitude or even stronger
than that of positive greenhouse gas forcing.
• There is a large spread of model-calculated
aerosol RF even in the global annual averaged
values. The AeroCom study shows that
the model diversity at some locations (mostly
East Asia and African biomass burning regions)
can reach ±3 W m-2, which is an order
of magnitude above the global averaged forcing
value of -0.22 W m-2. The large diversity
reflects the low level of current understanding
of aerosol radiative forcing, which is
compounded by uncertainties in emissions,
transport, transformation, removal, particle
size, and optical and microphysical (including
hygroscopic) properties.
Figure 3.3. Aerosol optical thickness and anthropogenic
shortwave all-sky radiative forcing from the AeroCom
study (Schulz et al., 2006). Shown in the figure: total AOD
(a) and anthropogenic AOD (b) at 550 nm, and radiative
forcing at TOA (c), atmospheric column (d), and surface
(e). Figures from the AeroCom image catalog (http://
nansen.ipsl.jussieu.fr/AEROCOM/data.html).
(a) Mean AOD 550 nm
(b) Anthropogenic AOD 550 nm
(c) Anthro. aerosol TOA forcing (W m-2)
(d) Anthro. aerosol atmospheric forcing (W m-2)
(e) Anthro. Aerosol surface forcing (W m-2)
In contrast to longlived
greenhouse
gases, anthropogenic
aerosol radiative
forcing exhibits significant
regional and
seasonal variations
edit on 2-3-2011 by MathiasAndrew because: (no reason given)