Deposition of ionic species and black carbon to the Arctic 1 snow pack : Combining snow pit observations with modeling 2

Although aerosols in the Arctic have multiple and complex impacts on the regional climate, their 16 removal due to deposition is still not well quantified. We combined meteorological, aerosol, precipitation, and 17 snow pack observations with simulations to derive information about the deposition of sea salt components and 18 black carbon (BC) from November 2011 to April 2012 to the Arctic snow pack at two locations close to Ny19 Ålesund, Svalbard. The dominating role of sea salt and the contribution of dust for the composition of 20 atmospheric aerosols were reflected in the seasonal composition of the snow pack. The strong alignment of the 21 concentrations of the major sea salt components in the aerosols, the precipitation, and the snow pack is linked to 22 the importance of wet deposition for the transfer from the atmosphere to the snow pack. This agreement was less 23 strong for monthly snow budgets and deposition indicating important relocation of the impurities inside the snow 24 pack after deposition. Wet deposition was less important for the transfer of nitrate, non sea salt-sulfate, and BC 25 to the snow during the winter period. The average BC concentration in the snow pack remains small with a 26 limited impact on snow albedo and melting. Nevertheless, the observations also indicate an important 27 redistribution of BC in the snowpack leading to layers with enhanced concentrations. The complex behavior of 28 bromide due to modifications during the sea salt aerosol formation and remobilization in the atmosphere and in 29 the snow were not resolved due to the lack of measurements in aerosols and precipitation. 30

D wet = C precip · P · 0.001 (2) 203 Major sea salt components and nitrate were determined using ion chromatography in precipitation samples 204 collected on a weekly basis using a bucket funnel system in summer and a snow sampler in winter (Kühnel et al.,205 2011; Aas et al., 2013). The data downloaded from the EBAS database (ebas.nilu.no) were used without further 206 correction, although the bulk sampler likely collected also gaseous compounds and particulate material due to 207 dry deposition. Especially in periods with high wind speed, the bulk collector may also catch large sea spray 208 aerosols. However, the exact contribution of dry deposition to the here calculated wet deposition is difficult to 209 quantify since it depends on the frequency of rain events and episodes with elevated sea salt aerosols. The total 210 and monthly wet deposition was calculated as the sum for the period from 31 October 2011 to 1 April 2012 and 211 for each month (except October). No measurements of bromide and BC in the precipitation are available. For 212 bromide, wet deposition was estimated from the wet deposition of sodium also applying the standard sea water 213 ratioscomposition according to D wet (bromide) = 0.00624 · D wet (sodium) (Millero et al., 2008). Wet deposition of 214 BC was estimated according to the scavenging scheme proposed by Sharma et al. (2013). The change in 215 atmospheric BC concentration ∆[BC] was estimated using the BC concentration [BC], the scavenging 216 coefficient (R; m 2 kg -1 ), the precipitation rate (P t , L m -2 s -1 ) and the time step (∆t, s) according to Eq. (3): 217 We used a scavenging coefficient of R = 5 · 10 -3 m 2 kg -1 as recommended by Sharma et al. (2013). Since all 219 scavenged atmospheric BC will be mixed into the accumulated weekly snowfall ((P t · ∆t), the BC concentration 220 in the snow [BC]  The stratigraphy and densities for the two analyzed snow pits are shown in Fig. 2. The investigated snow layers 232 comprised depths down to -263 cm below the surface for snow pit KV and -195 cm for snow pit AL reaching in 233 both cases the surface of the ice layer formed during the previous summer. While both snow pits showed a 234 typical increase in density from the surface to the deeper layers, the variability in terms of grain types and layer 235 structures was higher for snow pit AL. At an altitude of 670 m a.s.l. the high wind speeds at snow pit KV led to 236 the formation of several wind-packed layers. The impact of significant melting was not identified in the snow pit 237 KV, although the recorded temperatures reached several times values above or close to the melting point (see 238 Supplementary Material, Fig. S2). In contrast, at an altitude of 340 m a.s.l. melting events were more apparent in 239 snow pit AL, which exhibited several melt freeze crusts probably due to warmer periods in November 2011 and 240 January 2012 accompanied by air temperatures above 0°C and large amounts of rain at sea level in Ny-Ålesund. 241 The stronger impact of melting in the snow pit AL was confirmed by the chemical composition. The ratio of 242 magnesium to sodium has been proposed as a melt indicator (Iizuka et al., 2002;Virkkunen et al., 2007;Ginot et 243 al., 2010) with lower ratios caused by the preferential removal of magnesium due to percolating water. While the 244 average magnesium to sodium ratios were around 0.12 in both snow pits, in snow pit AL the variability was 245 higher and minimum values lower. Smallest ratios were encountered in layers deposited in November, January, 246 and March corresponding to the months with elevated air temperatures. Nevertheless, in both snow pits the ratios 247 did not reach the small ratios as observed in ice cores from Svalbard (Iizuka et al., 2002;Virkkunen et al., 2007). 248 Therefore, a certain redistribution of the impurities probably occurred in the snow pack due to melting with a 249 stronger impact on the Austre Lovenbreen glacier. This is, however, unlikely to have led to , but not a complete 250 elutionremoval. While the impact was stronger on the Austre Lovenbreen glacierHence, the overall budgets of 251 both snow pits are assumed to be mostly unaffected by melting.seemed not to be influenced. Ålesund was slightly higher than 30 % per 100 m altitude increase and is, thus, close to accumulation gradients 266 previously applied for the nearby Midre Lovénbreen and Austre Brøggerbreen glaciers (Hodson et al., 2005). linear decrease in accumulation from the snow pit KV to AL, monthly layers were also attributed to snow pit AL 277 with linearly interpolated depth ranges using the ratio of the total snow heights of both pits (Fig. 2). 278

Impurity profiles in the snow pack 279
Co-located impurity profiles were established for each measured compound combining the measured 280 concentrations with either the observed stratigraphy for both snow pits or the simulated stratigraphy for snow pit 281 KV. Profiles were established by assuming homogeneous concentrations for the identified snow layers and by 282 adjusting the closest observed concentrations to the vertical extent of the observed or simulated layers.

Wintertime snow budgets and deposition of ionic compounds 323
Total snow budgets of all measured compounds for the two snow pits were calculated using three different 324 approaches: (i) simple budgets were determined by multiplying the average concentrations by the total SWE; (ii) 325 adjusted budgets were calculated from the interpolated density profile shown in Fig. 2 and co-located 326 concentration profiles like in Fig. 3; (iii) for snow pit KV simulated budgets were obtained by combining the 327 simulated density profiles with simulated concentration profiles. All calculated budgets are summarized in Fig.  328 4, which also shows the observed wet deposition at Ny-Ålesund and the estimated total dry aerosol deposition

342
Due to differences smaller than 21% in their budgets, errors of the manual snow density measurements, the 343 chemical analysis, and the extrapolation of the density and concentration profiles, it can be assumed that 344 differences in the budgets below 20 % as obtained for chloride, sodium, magnesium, calcium, and potassium are 345 not significant. The spatial variability of snow concentrations at a scale of meters can be even larger (e.g. 346 Svensson et al., 2013). Thus, the total snow budgets for the pits KV and AL reveal a consistent picture for the 347 sea salt components chloride, sodium, magnesium, potassium, and bromide ( Fig. 4) with insignificant 348 differences in the observed total budgets despite differences in . For these species neither the method for the 349 calculation of the total budgets, the location, the altitude and, nor the accumulation led to significant differences 350 in the observed total budgets. This is consistent with recent observations revealing characteristic patterns of 351 aerosol concentrations along Svalbard glaciers including the Kongsvegen demonstrating consistent formation, 352 transport, and exchange processes between the atmosphere and the snow (Spolaor et al., 2017). 353 If post-depositional processes are negligible, the total snow budgets of the impurities correspond to the input due 354 to the sum of the wet and dry deposition. Based on the comparison of the total snow budgets with the observed 355 wet deposition, the estimated dry deposition isare evaluated for the different impurities. The total snow budgets 356 of chloride, sodium, magnesium, and potassium agree well with the observed wet deposition at Ny-Ålesund with 357 differences smaller than 20 % for the period from October 2011 to March/April 2012. However, the recorded 358 wet deposition also includes variable contributions from dry deposition since the precipitation samples were 359 collected with an open bucket instrument (Kühnel et al., 2011). Nevertheless, the estimated dry deposition 360 corresponds to less than 5 % of the wet deposition of chloride, sodium, and magnesium and reaches a maximum 361 of 14 % for potassium (Fig. 4). Subtracting the nss-sulfate from the total sulfate shows that the dry deposition of 362 sulfate with marine origin also corresponds to less than 5 % of the total wet deposition. Since the estimated dry 363 13 deposition is considered as an upper limit, it can be assumed that its contribution for the total snow budget on the 364 Kongsvegen and Austre Lovenbreen glaciers during the period November 2011 to April 2012 remained small for 365 chloride, sodium, magnesium, potassium, and sea salt sulfate. The estimated wet deposition for bromide based 366 on sodium concentrations and the standard sea water ratio leads to an overestimation of more than 40 % 367 compared to the observed bromide in the snow pack (Fig. 4). This demonstrates that sea salt bromide is 368 undergoing important modifications during the formation of sea salt aerosols, in the atmosphere, or after 369 deposition (see Sect. 3.7). 370 Like for the sea salt components, a good agreement between the KV snow budget of nitrate and nss-sulfate and 371 the total deposition during the period from October 2011 to April 2012 is found. For these two compounds the 372 observed wet deposition at Ny-Ålesund remains significantly below the snow budget, while the missing fractions 373 are largely compensated by the estimated dry deposition. For nitrate, the dry deposition is comparable to the wet 374 deposition, whereas for nss-sulfate dry deposition even dominates de snow budget. The adjusted budgets of the 375 snow pit AL show ~50 % less nitrate and ~40 % less nss-sulfate compared to KV (Fig. 4), which may be related 376 to the spatial variability of the dry deposition of the two species. 377

Wintertime snow budgets and deposition of BC 378
Regarding the snow budgets, the differences in the rBC profiles and t average rBC concentrations are partly 379 The anti-correlation between accumulation and the average BC concentrations in the two snow pits points to an 399 important contribution of dry deposition, which is in agreement with the estimated dry and wet deposition of BC. 400 The dry deposition of eBC (Fig. 4) derived with a deposition velocity of 0.1 cm s -1 corresponds to approximately 401 half of the observed rBC budget at KV and is, thus, somewhat higher than the contribution due to wet deposition. 402 Despite the overall uncertainty related to the simplified methods for the estimation of the BC deposition, the 403 difference between the snow budgets and the total deposition remain below 25 % for the KV and below 45 % for 404 the AL snow pit. This important contribution of dry deposition is in contrast to wet and total deposition reported 405 for Ny-Ålesund for the winter 2012/2013 based on rBC measurements in falling snow and in the snow pack 406 (Sinha et al., 2018). From these observations it was concluded that the dry deposition of rBC remained 407 negligible. However, the authors also reported rBC fluxes at 300 m altitude on the Broggerbreen glacier, which 408 were twice as high as in Ny-Ålesund. While Sinha et al. (2018) claim that this increase is mainly due to the 409 higher accumulation on the glacier, additional dry deposition at higher altitudes cannot be excluded. Moreover, 410 the potential contamination of the snow pack close to Ny-Ålesund due to local power generation or a potential 411 mismatch between the budgets of the falling snow and the snow pack due to the removal by blowing snow were 412 not considered. 413 Previous model studies have indicated that BC in the Arctic is primarily removed through wet deposition (e.g.

Comparison of monthly snow budgets and deposition 420
To derive a higher temporal resolution of the snow budgets monthly snow budgets were calculated from layers 421 deposited in each month between November 2011 and March 2012. The monthly budgets are further compared 422 to monthly wet and dry deposition. Each weekly wet deposition was attributed to the month with the largest 423 overlap in time to derive the monthly wet deposition, while the monthly dry deposition was calculated from the 424 daily dry deposition. Monthly total deposition was calculated as the sum of the corresponding wet and dry 425 deposition. Figure 5 shows as example the results for sodium, nitrate, and BC. Results similar to sodium were in 426 general obtained for the other sea salt components. The dominating role of wet deposition for sodium and other 427 sea salt components and the larger contribution of dry deposition for nitrate and BC are also reflected in the 428 monthly budgets. For the months with recorded precipitation at Ny-Ålesund, the wet deposition of sea salt 429 components largely dominates the total deposition. This is in contrast to nitrate and BC, which show several 430 monthly budgets with higher values for dry than wet deposition. 431 The generally good agreement between the total budgets of the two snow pits and the wet and dry deposition 432 Nevertheless, it appears that for compounds with a larger contribution of dry deposition the agreement between 441 snow budgets and total deposition is somewhat better like in the cases of nitrate and BC (Fig. 5).

474
Calcium shows a different behavior compared to the other major sea salt components with a significant 475 enrichment of calcium in the aerosols as well as in the precipitation, which results in also causes calcium-to-476 sodium ratios above standard sea water in a large number of snow samples (Fig. 6). Such an enrichment in the 477 Arctic may be attributed to calcium-rich aerosols originating from soils (Toom-Sauntry  precipitation, and in the snow do not exhibit a constant ratio compared to sodium (Fig. 6). On average, The 483 highest and lowest ratios are found in the aerosols and the lowest ratios in the precipitation with the average 484 snow pack ratio in between these values. This confirms that the nitrate and nss-sulfate in the snowpack can be 485 attributed to a mixture of wet deposition and dry deposition of aerosols. Although in wintertime the reactive 486 nitrogen budget is dominated by particulate nitrate (Hara et al., 1999), a further dry deposition of gas phase 487 species to the snow is possible, which may be even more important than the aerosol deposition (Björkman et al., 488

2013). 489
Due to the different sources of BC and sodium (long-range transport vs local or regional formation of sea salt 490 aerosols), no consistent BC-to-sodium variation is found in the aerosols (Supplementary material, Fig. S3). 491 Similarly, the variation of BC in the snow pits is also independent of the sodium concentrations. Since BC 492 particles are preferentially coated by organic matter or sulfate (Liu et al., 2011), atmospheric BC shows a linear 493 positive relationship to nss-sulfate resulting in a coefficient of determinationrrelation coefficient R 2 of 0.60 (Fig.  494   7). In the snowpack, the rBC-to-nss-sulfate ratios are less consistent and the average ratio is almost one order of 495 magnitude smaller than in the atmosphereaerosols. Despite the different measurement techniques for BC in the 496 aerosols and in the snow, the lower BC-to-nss-sulfate ratio in the snow can only partly be explained by the 497 different measurement methods. Different ratios in the snow may be caused by the smaller contribution of wet to 498 total deposition of BC as compared to nss-sulfate (Fig. 4). Moreover, the AL snow pit shows a higher variability 499 in the BC-to-nss-sulfate variation than the KV snow pit (Fig. 7) indicating that redistribution of the impurities 500 caused by melting probably also impacted BC and nss-sulfate.

Bromide in the snowpack 508
Since no bromide concentrations in the aerosols and in the precipitation are available, the ratio between bromide 509 and sodium is shown in the form of profiles for the snow pits KV and AL (Fig. 8). The ratio between the overall 510 bromide and sodium budgets varies between 0.0045 for KV and 0.005 for AL and is, thus, below the standard 511 sea water ratio of 0.00624 (Millero et al., 2008). Only distinct layers show enrichments of bromide (Fig. 8). 512 Multiple photochemical processes occur in the sea ice-snow-atmosphere system of the Arctic acting upon the 513 variation between bromide and sodium (Simpson et al., 2007;Jacobi et al., 2012). On solid surfaces (aerosols, 514 snow, sea ice) bromide can be transformed into volatile bromine compounds that are released to the atmosphere 515 and are subsequently deposited. Therefore, bromide can be depleted already in the sea salt aerosols generated 516 over sea ice, which would cause a wet and dry deposition flux lower than estimated based on the standard sea 517 water ration, or it can be diminished in the surface snow after deposition (Jacobi et al., 2012) explaining the 518 average bromide-to-sodium ratios below the sea water ratio in both snow pits. Nevertheless, since the released 519 bromide is subsequently deposited, a snow pack with layers enriched in bromide is also possible depending on 520 the dominating influence of the release vs the additional deposition of bromide (Simpson et al., 2007).  The chemical composition of aerosols, precipitation, and the snow pack was analyzed for Ny-Alesund, Svalbard. 550 The results concerning the snow budgets, the wet deposition, and the ratios of the different components in the 551 snow pack, in the precipitation, and in the aerosols underline the importance of wet deposition for the major sea- the estimated dry deposition remains well below 10 % of the total deposition for chloride, sodium, and 557 magnesium, while it contributes more than 20 % to the snow budget of calcium and potassium probably due to a 558 stronger dust contribution. It is possible that the relatively high overall accumulation including strong 559 precipitation events in the last week of January contributed to the high input due to wet deposition during the 560 winter 2011/2012. Therefore, the contribution of dry deposition of sea salt aerosols could be larger during winter 561 periods with different precipitation characteristics. Nevertheless, it appears that the wet deposition measurements 562 at Ny-Ålesund can be used to estimate the total wintertime deposition of the major sea salt components in the 563 areas surrounding Ny-Ålesund. 564 In contrast to the major sea salt components the dry deposition of nitrate and nss-sulfate was more important 565 than the wet deposition. However, the dry deposition of the corresponding gas phase species like HNO 3 and SO 2 566 are not well quantified (e.g. Zhang  Bromide is the sea salt compound showing the strongest variability in the ratio to other major components like 580 sodium, which is related to its high mobility in the sea ice-atmosphere-snow system caused by chemical 581 processes. Systematic measurements of bromide not only in the snowpack, but also in the aerosols, in the 582 precipitation, and in fresh snow are required to further investigate processes before the formation of the sea salt 583 aerosols, during their transport, or after the deposition to the snow pack. 584 While the annual budgets and estimated deposition for most of the studied species agree well, the results for the 585 monthly budgets obtained with the detailed snowpack modeling are less convincing. Further improvements 586 regarding the modeling of the Arctic snow pack are needed to better address physical properties (e.g. the 587 evolution of the snow density) and post-depositional processes acting upon the vertical distribution of impurities 588 in the snow pack. Although the treatment of impurities was recently implemented into the Crocus snowpack 589 model (Tuzet et al., 2017), the impact of processes modifying the vertical distribution of impurities in the Arctic 590 snowpack like blowing snow, sublimation, and percolation are still not fully considered in most models. The full 591 implementation of post-depositional processes into complex snow models may offer the opportunity to exploit 592 further snow pack and ice core observations for the reconstruction of climate and pollution. 593