Aerosol optical depth over the Arctic: a comparison of ECHAM-HAM and TM5 with ground-based, satellite and reanalysis data

We compare ground-based measurements of aerosol optical depth and ˚Angstr¨om parameter at six Arctic stations in the period 2001–2006 with the results from two global aerosol dynamics and transport models, ECHAM-HAM and TM5. Satellite measurements from MODIS and the MACC reanalysis product are used to examine the spatial 5 distribution and the seasonality of these parameters and to compare them with model results. We ﬁnd that both models provide a good reproduction of the ˚Angstr¨om parameter but signiﬁcantly underestimate the observed AOD values. We also explore the e ﬀ ects of changes in emissions, model resolution and the parametrization of wet scavenging.


Introduction
The Arctic is extremely vulnerable to past and future climate change, through complex interactions which can lead to severe regional impacts on the local hydrology, cryosphere and ecosystems, and to feedbacks on the global climate system (IPCC AR4, 2007). Significant changes have already occurred during recent decades, involv- 15 ing loss of sea-ice and snow-cover (e.g., Serreze et al., 2007) and affecting other important components of the environment (e.g., Post et al., 2009;Richter-Menge and Overland, 2011, and references therein). Anthropogenic and natural atmospheric aerosols play a crucial role in these processes. While concentrations in the Arctic are low on average, they reach a maximum in spring, forming the so-called Arctic haze, composed removal processes leads to a difficult identification of the transport pathways, which has advanced only recently (see, e.g., Koch and Hansen, 2005;Stohl, 2006;Shindell et al., 2008;Hirdman et al., 2010;Huang et al., 2010;Bourgeois and Bey, 2011).
Global atmospheric and aerosol models are important tools for studying climatic feedbacks and for estimating their impact on the climate system. They also provide 15 boundary conditions for higher resolution regional models. In both cases, a reasonable representation of the concentration and optical properties of aerosols over the Arctic is required. However, the verification of these models has been performed mainly on the global scale, focussing at low and mid latitudes, and their skill in the Arctic is still largely unexplored. Aerosol concentration measurements in the Arctic are sparse, 20 and the main source of data for model verification is provided by measurements of atmospheric optical properties collected by a network of a small number of measuring stations and by satellite observations.
In this work we compare the aerosol optical properties as modeled by two state-ofthe-art aerosol models with ground-based station measurements in the Arctic. We con-scavenging for HAM which has been recently suggested (Bourgeois and Bey, 2011).
In the following, section 2 provides information on the available station, satellite and reanalysis data and describes the ECHAM-HAM and TM5 models. In section 3 we report and discuss the comparison between observed and modeled aerosol optical properties. Concluding remarks are provided in Section 4. 15 2 Models and data

Ground-based measurements
We focus on long timeseries of sun photometer measurements of the daily mean values of AOD (500 nm) collected by groups participating in the POLAR-AOD programme (Aerosols Optical Depth in Polar regions; see Tomasi et al. (2007)  parameter α was derived by fitting a power law to spectra of AOD measured at different wavelengths. A complete description and analysis of these data is provided by Tomasi et al. (2007). In order to allow for a comparison with modeled, reanalysis and satellite AODs (which are all available at 550 nm), we interpolate measured AODs to 5 550 nm using the availableÅngström coefficients (the resulting values are lower by about 10% on average).

Satellite measurements and reanalysis
Satellite observations of AOD andÅngström parameter are provided by the Moderate Resolution Imaging Spectroradiometer (MODIS) aboard the Terra and Aqua satellites. 10 Specifically the Aerosol Cloud Water Vapor Ozone Daily L3 Global 1Deg CMG collection products were used. In this work we use data starting in 2001 (Terra) and in July 2002 (Aqua).Ångström parameters are based on AODs at 470 nm and 660 nm over the land and 550 nm and 865 nm over the ocean. We also consider an aerosol reanalysis product provided by the MACC (Monitoring 15 Atmospheric Composition and Climate) project (Benedetti et al., 2009), which uses the ECMWF IFS cycle 36R1 model with a prognostic aerosol scheme at resolution T255L60, assimilating MODIS AOD observations. Aerosol reanalysis data are available starting from 1 January 2003. To our knowledge a complete validation of the MACC reanalysis product in the Arctic is still missing. In this work we treat the MACC reanalysis 20 as an additional source of spatially extended AOD observations and we use it as an 'interpolator' to provide reference data also in months when satellite observations at high latitudes are scarce.

ECHAM5-HAM and emissions
The ECHAM5-HAM model couples the global climate model ECHAM5 (Roeckner et (2009)), which models the dynamics, the microphysics and the transport of the main atmospheric aerosols and their radiative feedbacks. In particular HAM contains the microphysical core M7 (Vignati et al. 2004), based on the representation of particle distributions as the superposition of log-normal modes peaked at different particle size classes, and reproduces the main aerosol emission, sedimentation and wet 5 and dry scavenging processes.  15 We integrate the ECHAM-HAM model using a "nudging" technique to force the model to stay close to the dynamical wind fields provided by the ECMWF ERA-Interim database (Dee et al., 2011) in the period 2000-2006. The nudging fields were prepared using the INTERA package (Ingo Kirchner, http://wekuw.met.fu-berlin.de/ ∼ IngoKirchner/nudging/nudging/). The initial year is used for spinup, and we consider 20 model results in the period 2001-2006. Since transport processes play an important role in determining the concentration of Arctic aerosols (Shindell et al., 2008), nudging was used in order to allow the model runs to reproduce as close as possible the main tropospheric winds in the period of interest and to allow a direct comparison with observed data over a limited range of years. 25 In this work we explore the use of different aerosol emission databases as boundary conditions for ECHAM-HAM, which are summarized in Table 1 together with the labels used to identify them. Table 2 details the total anthropogenic and wildfire emissions for these datasets, both globally and for the northern hemisphere excluding the equatorial 8324 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | area (north of 10 • N). In particular we use the AeroCom-I  and the ACCMIP (Atmospheric Chemistry and Climate Model Intercomparison Project, Lamarque et al., 2010) inventories, both for the year 2000. Anthropogenic fossil-fuel and bio-fuel emissions for sulfur, black and organic matter are annual data. Wildfire burning emissions are represented as monthly climatologies. Since fires in Siberia, Canada and Alaska have been found to provide a significant contribution to Arctic pollution (Law and Stohl, 2007;Stohl, 2006;Stocks, 1998;Koch and Hansen, 2005), with important interannual variability, the choice of biomass burning emissions datasets may play an important role. Variability in these emissions, interacting with interannual changes in circulation, may lead to significant changes in Arctic aerosol concentrations. While 10 ACCMIP includes data from the Global Fire Emission Database 2 (GFED2), we also explore ACCMIP emissions using the more recent GFED3 monthly biomass burning emission database (van der Werf et al., 2010) and we allow these emissions (sulfur, black carbon and organic matter) to vary annually instead of using a climatological mean.
Koch and Hansen (2005) showed that, while the largest contribution in terms of sulfate aerosols to Arctic AOD comes from Russia, South-East Asia contributes with a significant fraction. Since there is a significant difference in terms of sulfate emissions between the REAS and ACCMIP emission datasets, we explore the possible impact of a change in anthropogenic SO 2 emissions in the South-East Asia region. To this end,  In the following we also explore a simple change in model parametrizations suggested recently by Bourgeois and Bey (2011) to better reproduce the observed optical properties and concentrations of aerosols in the Arctic region. In particular they suggested to modify the wet aerosol scavenging parametrization in HAM, reducing the corresponding scavenging coefficients and thus increasing aerosol lifetimes. We apply 5 this modification using the same parameters as described in their paper.

TM5
TM5 is a global three-dimensional atmospheric chemistry and transport model (Krol et al., 2005). Aerosol microphysical processes are modeled using the aerosol dynamics module M7 (Vignati et al., 2004). It represents sulphate, black carbon, organic carbon, 10 sea salt and mineral dust in seven internally mixed soluble or insoluble log-normal size modes. Gas-particle partitioning of ammonium nitrate is calculated using the EQSAM thermodynamic equilibrium model, as described in Aan den Brugh et al. (2011). The gas-phase chemistry scheme is based on the Carbon Bond Mechanism 4 (CMB4) and is given in Huijnen et al. (2010). In our setting, the horizontal resolution is 3 × 2 degrees 15 and the vertical grid comprises 34 hybrid σ-pressure levels. Atmospheric dynamical fields are provided by ECMWF ERA-Interim reanalysis data (Dee et al., 2011).
Anthropogenic and biomass burning emissions are taken from the CMIP5/ACCMIP datasets. The year-2000 values from the historical dataset described in Lamarque et al. (2010) are combined with scenario estimates for the year 2005 and 2010 from the 20 representative concentration pathway RCP4. 5 (van Vuuren et al., 2011). Linear interpolation is applied for the intermediate years.
Emissions of sea salt, oceanic DMS and nitrogen oxides (NOx) from lightning are calculated online, while other natural emissions are prescribed. Sea-salt emissions are parameterized as described in Vignati et al. (2010). The schemes applied for DMS and 25 lightning NOx are the same as in Huijnen et al. (2010). The emissions of mineral dust are prescribed using the AeroCom-I dataset for the year 2000   . To indicate the data availability and interannual variability we report the individual monthly means of the station data for every year, instead of their average. The figure also reports the results from the MACC reanalysis and the MODIS Terra and Aqua climatologies. At some sites (NyÅlesund, Sodankyla and Barrow) ground-based observations, MODIS and MACC data display similar values. For these sites we observe 20 that the MACC dataset agrees well with the observations, also in terms of seasonality. A significant late spring-early summer maximum of AOD is clearly visible, with maximum average AOD values exceeding 0.1. At these sites, both ECHAM-HAM and TM5 produce values of AOD which are significantly lower than the observations, particularly during the spring maximum. Only at the rather meridional station of Sodankyla Introduction Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | peak in July instead of March-April as for the observations and the MACC reanalysis. The other Scandinavian station, ALOMAR, shows larger differences between station data and satellite data, particularly in spring. In this particular case though, the agreement between station data and modeled AOD is good, particularly for ECHAM-HAM. At Alert the modeled AOD is significantly smaller than station data, MODIS and MACC 5 observations (there are some significant differences also between these datasets, presumably for the lack of reliable satellite data in this area). At Summit, MACC estimates are very different from observed data during the whole year, presumably because there are no MODIS satellite data available for this location. Modeled AODs are significantly lower than both ground-based measurements and the MACC reanalysis. To summarize, the models underestimate AOD, except at Sodankyla and ALOMAR, and they peak in summer, while the observations, generally, show a maximum in spring (as documented also in Shaw (1995)). It has been suggested in Bourgeois and Bey (2011) that the amount of aerosols transported to the Arctic is sensitive to the magnitude of wet removal, suggesting a 15 revision of the wet scavenging parameters used in the ECHAM-HAM model. We tested this approach, using the parameter values suggested in that paper, and the resulting AOD climatology is shown as a blue line in Fig. 3. We see a significant increase of AOD at all latitudes, leading to a reasonable agreement of summer AOD values with observations at all sites. Like the observations, the AOD modeled with the Bourgeois 20 and Bey (2011) parametrization displays a peak in spring, but instead of presenting a minimum in winter as suggested by MACC and by the available measurements, it is minimum in summer.

Timeseries and monthly climatologies
The monthly climatology of theÅngström parameter is reported in Fig. 3b. While the exact values of theÅngström parameter shown here can be compared only with 25 difficulty, as they were all obtained with slightly different methods (a caveat discussed in Tomasi et al. (2007)), comparing their seasonal variations is certainly of interest. While the observed AODs have been found to peak around April-May, theÅngström parameter is found to peak in June-July in the measurements, suggesting that there is To summarize, theÅngström parameter is reasonably reproduced at most stations by both models, indicating that the distribution of particle sizes is captured correctly, together with its seasonality, characterized by a peak in summer. The main exception is the Summit station, located at very high altitude on the Greenland ice 15 sheet, where both TM5 and ECHAM-HAM display an excess of fine particles.

Spatial distributions
The spatial distributions of the AOD climatologies are reported in Figs. 4, 5 and 6, all averaged over the years 2003-2006, which are common to all datasets. Panels 4a and b report the values of the AOD from MODIS data (Terra) and MACC reanalysis.

20
MODIS data are missing from November to February. In the other months, the MACC reanalysis and MODIS data display a similar spatial structure. Some differences are evident, such as the MACC AODs being lower than reported by MODIS in Scandinavia in July-August and higher over the Atlantic in July-October; nonetheless there is a general agreement between the two datasets, both in spatial distributions and amplitudes, 25 suggesting that the MACC reanalysis provides a reasonable interpolation of available observations in these areas. Both the satellite and the reanalysis datasets are characterized by a large spatial variability in AOD, with significant differences between the 8329 Introduction Atlantic and the Pacific sector and an average decrease of AOD with latitude. MACC predicts low AOD values from November to February, which, while they cannot be verified with MODIS observations, find some confirmation from station data (see fig. 3). For these reasons, in the following we are encouraged to use the MACC reanalysis as a reference to compare with modeled AOD distributions.
5 Figure 5 reports the AOD climatologies observed for ECHAM (EIPCC emissions) and TM5. A severe underestimation of AOD over the entire Arctic area and in all months is apparent for both models. A high AOD tongue over the Atlantic in all months is reproduced, particularly in ECHAM-HAM, but its amplitude is too weak when compared with MACC and with MODIS from March to June. This suggests that sea-salt aerosol emissions are possibly well reproduced but that the AOD associated with other components is underestimated. Spatial maps of the AOD due to only sea-salt (not shown) show an intense contribution of the Atlantic which peaks in December-January and confirm this conclusion. The contribution of the Atlantic in winter is much weaker in TM5 than in ECHAM-HAM possibly due to differences in the emission parametrization. Increased 15 values of AOD over northern America and Russia in May-August, possibly associated mainly with fires, are present as spatial features in both model runs, but underestimated in amplitude. Spatial maps of the contribution of organic carbon (not shown) confirm this view showing contributions over Siberia and Northern Canada in these months. Overall, as already seen for the monthly climatologies at the station locations, the modeled data do not show the observed seasonality of Arctic AOD, characterized by a pronounced peak in late spring and early summer. Average amplitudes over large areas are underestimated by almost one order of magnitude in some months.
Introduction of the Bourgeois and Bey (2011) modification, in Fig. 6a, leads in general to larger amplitudes which are, in all seasons, closer to the observed (MODIS) and served in late winter/spring are still underestimated. Overall the seasonal variability seems lower than observed, compare for example the change between July-August and September-November with the MODIS observations.
Previous works (Law and Stohl, 2007;Stohl, 2006;Stocks, 1998) suggest that fires in spring and summer in these regions can be the dominant source of sulfur and black 10 and organic carbon. In Fig. 6b we report also results from the ECHAM-HAM model using the EGFED emissions, which include a better representation of fires in these regions and interannual variability. The overall picture does not change much, even if some patches of higher AOD can now be found on the continental masses of North America and Russia in summer.

Sensitivity to emissions, resolution and nudging
We further explore the ECHAM-HAM model sensitivity to changes in the emission databases in Fig. 7, which reports, for the different emissions, the monthly AOD climatologies at the measurements sites for which ground-based data are available. We find clearly that, while these emission databases differ significantly in terms of annual 20 average emissions (see the budgets reported in Table 1), their impact on Arctic AOD is scarcely significant in the model simulations, irrespective of whether or not the Bourgeois and Bey (2011) modification is used. Neither a better representation of Arctic fire emissions (EGFED), including interannual variability, nor the increased emissions in Southeast Asia, do impact the modeled Arctic AOD. In principle it is possible that 25 interannual variations in atmospheric circulation may interact with the timing of major fires in the Arctic, leading to differences in summer aerosol concentrations. Our results indicate that this is not the case. A simulation (not shown) using a monthly climatology 8331 Introduction based on the EGFED emissions leads to results which are almost indistinguishable from the EGFED simulation with interannual variations. For comparison we also report the results obtained (for the same years) from a climatological, non-nudged ECHAM-HAM simulation using the EAERO emissions. The results display some small differences with respect to the nudged runs, particularly at 5 the European stations of NyÅlesund, ALOMAR and Sodankyla, in keeping with the view that the details of the atmospheric circulation and of the associated transport processes influence the aerosol distribution. However, the differences between the nudged and non-nudged runs are rather small, and in all cases the simulated AOD remains smaller than the observations at all sites. 10 The role of model resolution is also explored, considering a nudged ECHAM-HAM simulation (with EAERO emissions) at a higher horizontal and vertical resolution (T63 with 31 vertical levels), also reported in Fig. 7, averaged for the years 2001 to 2003. We find that, while there are some differences in the average AOD observed in individual months (possibly due to the shorter period over which the climatology has been 15 computed), these do not cure in any way the underestimation of the AOD which we already discussed. This result is in agreement with Shindell et al. (2008), where it has been shown that, in an intermodel comparison, horizontal model resolution does not show any clear effect on aerosol transport towards the Arctic.

20
To get an insight into the contribution of individual aerosol compounds, Fig. 8a-d report the average over the entire Arctic (defined as the area north of 60 • N) of the total AOD and of the fraction associated with each aerosol compound (organic carbon, black carbon, sea salt, sulfur and dust). For MACC and TM5, ambient aerosol contributions to total AOD are reported (i.e. the contribution of their water content is included) while 25 for ECHAM-HAM we report dry compound contributions and diagnosed aerosol water as a separate component. As the MACC results (Fig. 8a)  it. An "Arctic-haze" spring maximum in April-May is clearly visible, with a progressive decay during summer, reaching a minimum in December. Sea salt has an opposite seasonality, reaching a minimum in summer and a maximum in winter, compatible with a contribution of Atlantic sea salt emissions in winter which we have described above, possibly due to stronger mesoscale perturbations. Organic carbon peaks and 5 contributes significantly to total AOD from May to August. As already mentioned, spatial maps of its distribution (not shown) confirm an association with wildfires in Siberia and norther Canada in that period. Also black carbon peaks in summer, but shows a secondary peak in the Arctic haze months of April and May. There is also a significant contribution from dust in the MACC reanalysis. We can compare these results with the 10 AODs attributed to individual compounds in the TM5 simulation in Fig. 8b (note the different vertical scale). For reference, in this figure we also report AOD due to water in aerosols, even if the shown AOD fractions are for ambient aerosols. We see that, as already discussed, the total average AOD is underestimated by a factor of more than four and there are significant differences in seasonality and variability. Sulfates still play 15 the dominant role, accounting for almost half of the observed total AOD in summer, but instead of reaching a maximum in April-May, they peak in summer, from June to August. Like the other compounds their AOD is significantly smaller than reported by MACC. Organic and black carbon peak in July and August, showing a seasonality similar to MACC, which can be understood in terms of seasonality of wildfires (as confirmed 20 by spatial maps of its distribution, not shown). The contribution of dust in TM5 is significantly lower than in MACC. Sea salt shows the same seasonality as in MACC but also these values are lower by about a factor of three. Nitrates provide a very small contribution to total AOD, comparable to that of black carbon in winter. The aerosol water contribution to total AOD, shown in the figure, appears to be dominated mainly by 25 the the seasonality of sulfates, with a contribution by sea-salt in winter. These effects in terms of AOD are associated with corresponding seasonalities in terms of total Arctic loads (burdens) which we report in Fig. 8f Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | a significant dust load is present, we have seen above that it does not contribute much to the total AOD. Overall the partition in terms of the contribution of individual aerosol compounds to total AOD is similar between TM5 and MACC, leading to the conclusion that (provided the MACC reanalysis is reliable), the underestimation of total AOD may be due to mechanisms which affect all compounds, such as errors in long range 5 transport or underestimated aerosol residence times for this model. The AOD contributions of individual compounds for the ECHAM-HAM simulations (using EIPCC emissions) are reported in Fig. 8c and the corresponding Arctic burdens are reported in Fig. 8e. In this case, since aerosol water content is treated as an additional optically active compound in HAM and its contribution is saved separately in 10 the diagnostics, we report the dry contributions to the total AOD of each compound and, separately, the contribution of diagnosed aerosol water. The total AOD, while on average comparable to that reported by TM5 and much smaller than the MACC average, presents a smaller seasonal variability compared to MACC. Also in this case AOD peaks in August, while high values in April are absent. Sulfates follow the same 15 pattern, with a weak summer maximum and no sign of a spring peak. We find again that organic and black carbon are concentrated in summer months. The total burdens, reported in Fig. 8e are, compared to TM5, lower in terms of sulfates, black carbon and dust and similar for organic carbon and sea salt.
If we introduce the Bourgeois and Bey (2011) wet scavenging modification, total 20 AODs change significantly, as reported in Fig. 8d. The average value is now comparable to what reported by MACC, but with a very different seasonality. The peak in AOD has now shifted to February-March, while July corresponds to a minimum. Aerosol water and sulfates follow the same pattern. Organic carbon still peaks in summer, while black carbon AOD and concentrations (not shown) reach now a peak in winter and a 25 minimum in summer. Overall the wet scavenging modification, while leading to more realistic values of Arctic AOD as a time average, leads to a different seasonality in which a summer minimum in AOD appears, instead of a minimum in winter months as indicated by the MACC reanalysis. This leads to significantly higher AOD estimates in 8334 winter and very low values in summer. The seasonal variability in total AOD produced by ECHAM-HAM with the wet scavenging modification remains smaller than indicated by MACC.

Discussion and conclusions
An underestimation of modeled concentrations of sulfates and black carbon in the Arc- 5 tic has already been evidenced in Shindell et al. (2008) for several current models, including TM5 and ECHAM-HAMMOZ (ECHAM-HAM + gas phase chemistry module MOZART; Pozzoli et al. (2008)). Comparison with a large number of other models in that study has allowed to determine a great diversity in model results, attributed mainly to differences in aerosol physical and chemical processing mechanisms, while emis- tic. In comparison the model is successful in reproducing long-range transport of CO, suggesting that dry transport processes are reasonably well modeled. They suggested a reduction of the wet scavenging parameters in the model, finding significant improvement.
In this work, we explored the skill of TM5 and ECHAM-HAM in these areas in 25 deeper detail, focusing on the optical parameters AOD andÅngström, which are directly observed by ground-based and satellite measurements, comparing both their Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | spatial structure and the amplitudes measured at the location of six Arctic stations. TheÅngström parameter is reasonably reproduced at most stations by both models, indicating that the distribution of particle sizes is captured correctly, together with its seasonality, characterized by a peak in summer. The main exception is the Summit station where both TM5 and ECHAM-HAM overestimate the parameter; this station 5 is located at high altitude on the Greenland ice sheet, so that orographic effects and model resolution may play a role. The AOD results confirm a severe underestimation in amplitude of observed values and an absence of the strong seasonality found in the observations, with a peak in summer rather than in late spring. Spatial maps show that some observed features, such as the appearance of areas with very high AOD 10 over north America and northern Russia from May to August, are not reproduced at all in the model simulations. We verified, for ECHAM-HAM, that changes in emission databases or in model resolution do not have a significant impact on the Arctic distributions of modeled AOD. There are also no significant differences if the model is free to run in a climatological mode or forced by nudging to follow observed winds. These 15 results suggest that, at least for ECHAM-HAM, mechanisms different from transport, such as aerosol physical and chemical processes are more important to explain the model deficiencies, in agreement with the conclusions of Shindell et al. (2008). When we test the modification of wet scavenging suggested by Bourgeois and Bey (2011) we find much higher AODs, comparable in a yearly and spatially averaged sense with 20 observations, but still with very different seasonality and spatial structure. In particular, values in autumn and winter appear too high, while areas with high AOD over northern Russia and America in summer are not reproduced. It is important to notice that a comparison between aerosol measurements and model results in the Arctic area is particularly difficult. Indeed, aerosol measurements are particularly challenging and Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | From the results reported above we conclude that efforts towards a better representation of aerosols in the Arctic, for ECHAM-HAM and possibly TM5, should be mainly aimed at improving current parametrizations involving aerosol removal and transformation. As we have seen, model results are only weakly sensitive to changes in emission databases and transport. On the other hand, changes in wet scavenging, while they do not produce a correct seasonality and spatial aerosol distribution, allow to reach yearly averaged values of AOD in the Arctic which are more realistic. Several improvements which may be beneficial are being developed and need to be investigated in detail. A recent study (Browse et al., 2012) associates the seasonal cycle of Arctic aerosols with scavenging processes, in particular with the passage from inefficient scavenging of soluble aerosols from ice clouds in winter to more efficient scavenging from low, warm liquid clouds in summer. Croft et al. (2010) compared different in-cloud aerosol scavenging parametrization schemes for ECHAM-HAM, finding large variabilities in aerosol mass and number burdens, and improved agreement with recent diagnostic and prognostic scavenging schemes, particularly at high latitudes. Improved schemes for below-15 cloud scavenging by snow in ECHAM-HAM are discussed in Croft et al. (2009).
We conclude noting that the spatial maps of observed AOD, from satellites and reanalysis, show a large spatial variability of this parameter over the Arctic, suggesting that measurements at single stations can be hardly representative of larger areas. On the other hand, ground-based measurements are essential to calibrate and validate 20 satellite measurements and reanalysis datasets. Due to the impact of changes in the Arctic for the global climate, the extensions of the currently available measurement network, the collection and availability of long-term measurements and the exploration of methods to bound data reliability are important future goals.

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version