Ancient versus modern mineral dust transported to high-altitude Alpine glaciers evidences Saharan sources and atmospheric circulation changes

Introduction Conclusions References

tica ice cores reveal higher continental dust deposition during the 20th-century warming (De Angelis and Gaudichet, 1991;Thompson et al., 2000;McConnell et al., 2007;Thevenon et al., 2009).Moreover, despite the striking increase of mineral dust transported over Europe during the last decades and the associated radiative forcing on climate system (IPCC, 2007), there is a lack of ice-core data about mid-latitude dust characteristics covering the preindustrial period.In order to evaluate European preindustrial atmospheric dust emissions, we have characterized windblown mineral dust archived in an Alpine ice core over the last millennium.In fact, mineralogical and geochemical compositions of the mineral aerosols entrapped in the ice differ according to their geological sources (Grousset and Biscaye, 2005).For instance, strontium (Sr) and neodymium (Nd) isotopic measurements of aeolian dust transported to East Antarctica drilling sites have highlighted efficient transfer of dust from South America during cold Quaternary climates (Delmonte et al., 2010).Similarly, Eastern Asia desert areas were identified as being the main source of dust found in Greenland ice cores from 44 to 14 kyr BP (Svensson et al., 2000).Nevertheless, ice-core records also reveal complex spatial and temporal patterns of variability that require further investigations (Delmonte et al., 2004;Xu et al., 2010;Vallelonga et al., 2010).In order to link aeolian dust sources and associated circulation patterns, we have studied the insoluble dust fraction transported to the Alps during the last millennium.Mineralogical and geochemical fingerprints of the paleodust were then compared to analog aerosols and potential dust source areas.
The Sahara is the world's major source of mineral dust, which subsequently spreads across the Mediterranean and Caribbean seas into northern South America, Central America, North America, and Europe.Grousset and Biscaye (2005)  over the Alps has been recovered (Jenk et al., 2006), makes a valuable candidate for studying past atmospheric dust sources and associated transport patterns.The objective of this study is 1) to characterize the mineralogy, geochemistry and isotopic composition (Sr and Nd) of aeolian dust windborne over the Alps during the millennium, and 2) to compare the paleodust characteristics to analog aerosols collected at the Jungfraujoch high-alpine research station and with documented Saharan dust sources.

Study sites
This study was performed with the samples collected from two sites: i) the Colle Gnifetti glacier saddle (CG, 45 • 55 N, 7 • 52 E; 4455 m a.s.l.; Fig. 1) is located in the Monte Rosa Massif (the second highest mountain in the Alps), on the southern side of the Alps and on the border between Switzerland and Italy.A previous study has demonstrated that the CG archive (core CG03) allows the reconstruction of changes in the dynamic of the southwesterly dust-laden winds from the Sahara, in relation to variability in large-scale atmospheric circulation patterns (Thevenon et al., 2009).The CG glacier saddle is located at ca. 70 km from ii) the high-alpine research station Jungfraujoch (JFJ, 46 • 33 N, 7 • 59 E; 3580 m a.s.l.; Fig. 2), which is the highest permanently manned weather station in Europe.This high Alpine site is located on the northern edge of the Swiss Alps, offering a very low background clean continental site for studying changes in the composition of the atmosphere (Nyeki et al., 1997;Overton, 2008).Daily PM 10 aerosol samples were collected on quartz fiber filters (24 h from midnight to midnight) within the scope of the Swiss National Air Pollution Monitoring Network (NABEL).Figures

Back Close
Full 3 Analytical methods

Dust samples and digestion procedure
The CG ice-core dust samples were extracted from cellulose membrane filters mounted on smear slides using Canada balsam, previously analyzed for total aerosols and mineral grain size by image analysis, using the procedure as described in Thevenon et al. (2009).Retrieved filters encompassing insoluble particulate matter were cut in two parts, and placed in 1.5 ml of xyl ène in an ultrasonic bath for 10 min to dissolve the Canada balsam.Afterward, the original filter was removed, and 5 ml of Milli-Q water was added in each sample.The first half of the ice core sample was filtered on Ag filter for being analyzed by X-ray diffraction, whereas the second part was digested in 1 ml HNO 3 (suprapur, 65%), 2 ml of HF (suprapur, 40%), and 1 ml of HClO 4 (suprapur, 70%).Following evaporation at 150 • C, 1 ml of Milli-Q water and 1 ml of HNO 3 (suprapur, 65%) were successively added, and the solution was left to complete evaporation between each step.The resulting solid was finally dissolved in 8 ml of 1% HNO 3 solution for chemical analysis.About 2 ml was used for elemental and REE analysis by ICPMS, while the remainder was further analyzed for isotopic composition by mass spectrometry as described in the Sect.3.4.Twelve ambient daily PM 10 aerosols samples enriched in atmospheric dust and collected between 2008 and 2009 at the high-altitude research station JFJ were selected.The quartz fiber filters (Whatman QMA and Pallflex Tissuquartz) were placed in an ultrasonic bath for 10 min.and prepared following the method described above.Gravimetric filter measurements (PM 10 ) were done by the Swiss National Air Pollution Monitoring Network.When sufficient dust material was present, the mean grain-size was measured with a laser granulometer (Malvern Mastersizer) by laser diffraction, after removing the organic fraction using H 2 O 2 treatment.Introduction

Conclusions References
Tables Figures

Back Close
Full

Mineralogical analysis by XRD
Bulk dust and clay mineral analyses were performed on Ag filter, with a X'TRA-ARL (Thermoscientific) diffractometer, using a wavelength of λ=1.540562Å.The peak areas of the clays were identified (2θ) with a WinXRXD profile-fitting program, using a Pearson-7 deconvolution function.Percentages of chlorite (C004 peak at 25.1 • 2θ Cukalpha reflection), mica (M001 peak at 8.88 • ), and kaolinite (K002 peak at 24.9 • ), were estimated using the following equations:

Trace and REE analysis by inductively coupled plasma mass spectrometry (ICP-MS)
The concentration of trace and REE elements (Sc, Ti, Cs, Ba, Hf, Ta, Pb, Th, U, La) in the digested solution was measured using quadrupole-based inductively coupled plasma mass spectrometry (ICP-MS) (HP 4500, Agilent).Total variation coefficients of three replicates sample measurements were smaller than 10%.

Isotopic analysis by Thermal Ionization Mass Spectrometry (TRITON) analysis
Sr and Nd separation from the solutions prepared as above described was carried out using cascade columns with Sr-spec, TRU-spec and Ln-spec resins following a modified method after Pin et al. (1994).Sr and Nd isotope ratios were measured on a Thermo TRITON mass spectrometer on Faraday cups in static mode using the virtual amplifier mode to eliminate cross-calibration effects on the amplifiers.Sr was loaded on single Re filaments with a Ta oxide solution and measured at a pyrometer-controlled 20172 Figures

Back Close
Full

Results and discussion
Trace element concentrations (Pb, Ti, Ba, Cs, U, and La) were normalized to conservative crustal elements (Sc, Ta, Hf, or Th) which primarily derive from wind-borne soil and rock-dust sources, and expressed in the form of crustal enrichment factors (EFs) after normalization to the mean concentration ratios in the upper continental crust (Wedepohl, 1995).EFs were very low (EF<2) for Ti, Ba, and La, thereby excluding important anthropogenic-induced fluxes of crustal elements (i.e.enhanced dust due to deforestation and agricultural activities).By contrast, Pb, and to a lesser extent Cs and U presented higher EFs (EF>10), suggesting the impact of atmospheric emissions of trace elements from anthropogenic sources (e.g.mining and open pit operations, smelting, nuclear tests).
Variations in crustal element abundances are primarily influenced by total and mineral dust inputs (Fig. 2).Dusty layers exhibit higher trace element enrichments and larger particles (>1.5 µm), whereas the light-coloured layers dust are depleted in heavy and/or incompatible elements and contain smaller particles.Such a distribution pattern suggests the activation of a predominant Saharan-derived dust source (dusty layers) 20173 Introduction

Conclusions References
Tables Figures

Back Close
Full in conjunction with the background signal recorded during low dust flux (light-coloured layers), rather than distinct contributions of local bedrock (Evans et al., 2004;Moreno et al., 2006).Moreover, the geochemical and mineralogical composition of the dust deposited at CG during the last centuries are quite similar to the composition of the dust collected in 2008 and 2009 at JFJ (Figs. 2, 3, 4 and 5), suggesting an overall similarity in aeolian dust sources through time.
Mineral characterization of CG and JFJ samples reveals strong similarities between modern and ancient dust composition, as well as the ubiquity of three source markers for Saharan dust that are palygorskite, kaolinite/chlorite ratio, and quartz (Table 1) (Glaccum and Prospero, 1980;Avila et al., 1997;Caquineau et al., 2002;Fiol et al., 2005).The clay mineralogy of CG and JFJ is reported on Table 2 and compared on Figure 5, showing rather similar clay composition.The dominant mineral group is micamuscovite, although occasional contributions of local bedrock might occur.This may be the case, for instance, of the great abundance of mica accompanying the presence of paragonite (a mineral present in the Southern Alps but rare to absent in North Africa) within the end of the greatest Saharan dust about 1790 (Fig. 4).Potassium feldspar and plagioclase were accessorily found in some of the ice-core and analog samples (Fig. 4).
The Sr and Nd isotopic similarity between CG ice core data and the available literature data (Grousset et al., 1998;Grousset and Biscaye, 2005) is remarkable, demonstrating the overall Saharan origin for the paleodust reaching the Alps (Fig. 3).This is particularly evident for the dusty ice-core layers (>100 mm 2 of dust/kg) which contain the larger particles (>1.5 µm), and for the sample covering one century of the Middle Age (1200-1300).However, less radiogenic Sr ratios are observed within the light-coloured layers (<30 mm 2 of dust/kg) or when the impact of the fossil fuel pollution becomes important, as attested by minima of 87 Sr/ 86 Sr values around 1910, 1935 and 1970, which coincide with maxima anthropogenic Pb deposition at CG (Figs. 2  and 3) and with the greatest lead emissions over Europe (Moor et al., 1996;Weiss et al., 1999).EF Pb reached ca.375 times background level around 1920, when anthro-Figures

Back Close
Full pogenic Pb abruptly increased because of the introduction of leaded petrol (Nriagu et al., 1990).The recent less radiogenic Sr ratios therefore reflect mid-latitude intercontinental dust background properties (i.e., physical weathering processes) and/or anthropogenic contamination (e.g., coal for power and smelting plants, traffic exhausts), rather than an alternative contribution from a deeply chemically weathered source (Lahd Geagea et al., 2008;Moreno et al., 2006).Figure 6 plots the aerosol surface, and the mica and chlorite contents against 87 Sr/ 86 Sr values to illustrate changes in the Sr radiogenic composition with increasing aerosol deposition and clay characteristics.
Higher 87 Sr/ 86 Sr ratios are found within the dusty layers and also show fair positive correlations with mica content and negative ones with chlorite content, probably due to the geochemical affinity of Rb with K, which is abundant in mica-muscovite but absent in chlorite (Fig. 6).Such features may also explain the fact that the JFJ samples are relatively depleted in radiogenic Sr with respect to CG samples (Fig. 3), since the accumulation time is very low.Air-mass back-trajectories were calculated for analogs using the Hysplit model (Draxler and Rolph, 2003) in order to identify the source region of the paleodust reaching high-altitude Alpine areas (Collaud Coen et al., 2004).The 315 h back-trajectories reaching JFJ are reported on Fig. 1 other JFJ samples (open squares on Fig. 2).Such results therefore support our assumption of Saharan contribution not only for the strong dust events that were rapidly transported towards the Alps from southerly directions, but also for some of the longrange intercontinental dust transport reaching the Alps from northerly directions.Previous air mass back-trajectories analysis from dust deposition sites in the Alps evidenced Saharan mineral dust but also China loess, transported during several days along a pathway across the eastern North Atlantic and approaching the Alps from northerly direction (Grousset and Biscaye, 2005;Sodemann et al., 2006).CG isotopic data furthermore suggest that intercontinental Saharan dust sources might contaminate glacial ice at summit Greenland and that Asian dust (e.g., Gobi desert) might reach the Alpine summits after long-range intercontinental transport within northwesterly winds (Burton et al., 2006;Svensson et al., 2000) (Fig. 3).
Long-term variations of Saharan dust transport to Alpine glaciers reflect atmospheric circulation changes over the Northern Hemisphere (Kang et al., 2003).Higher dust deposition at CG as evidenced during the first part of the Little Ice Age (LIA) (i.e., before ca.1690), within the greatest event of the millennium (ca.1780-1790), and after 1870, therefore strongly suggests winter drought conditions over North Africa and/or stronger spring/summer pressure gradient over the North Atlantic (Chiapello and Moulin, 2002).By contrast, low dust deposition during the latter part of LIA (from ca.1690 to 1870), reflects weakened pressure gradient across the North Atlantic region and increased rainfall in northern Africa (Glueck and Stockton, 2001;Thevenon et al., 2009).Weaker summer southwesterlies trade winds inferred from CG dust record during the latter part of the LIA, are consistent with the synchronous decline of the meridional overturning circulation (MOC) and increased negative states of the North Atlantic Oscillation (NAO) (Shindell et al., 2001;Lund et al., 2006).
Post-1850 increase in carbonaceous residues of combustion, or black carbon (BC), and in trace elements content, likely highlighted additional human impact on recent insoluble pollutant emissions (Fig. 2).The isotopic composition of the soot carbon (δ 13 C BC ) reflects dominant C 3 (δ 13 C=-27‰) woody pyrogenic emissions or fossil coal 20176 Introduction

Conclusions References
Tables Figures

Back Close
Full burning, while the decreasing trend observed in the δ 13 C BC during the last century likely reflects increasing relative contribution from fossil fuel emissions ( 13 C Suess effect; Thevenon et al., 2009) (Fig. 2).Polycyclic Aromatic Hydrocarbons (PAHs) analysis from CG core also reveals exponentially increasing PAHs emissions from 1900, reaching maxima concentration levels in 1920and 1940-1950(Gabrieli, 2008)).The present record, which ends with the 1977 Saharan dust event, therefore agrees with a strong decrease of anthropogenic pollutant emissions in the 1960s-1970s (Pb and U on Fig. 2), thanks to more efficient emissions control.Nonetheless, PAHs data furthermore show a renewed increase of anthropogenic pollutant emissions from 1975 to 2003 (Gabrieli, 2008).Additional data with a large spatial representation from shallow snow cores would therefore be necessary to assess seasonal changes of natural and anthropogenic trace elements deposition, with respect to sources and atmospheric pathways.

Conclusions
Mineral dust characterization reveals a dominant Saharan-derived source for the dust transported by southwesterly winds, and a subordinate background component linked to northwesterly winds, which becomes evident during periods of low dust deposition.Lower dust inputs in the ice core are depleted in crustal elements and in radiogenic Sr, supporting the occurrence of a background dust source and/or recent pollution contamination, rather than the activation of a secondary dust source.Mineralogical composition and Sr and Nd isotopic composition furthermore suggest that the Saharan intercontinental dust source is not only transported to the top of the Alps but also occasionally reaches Summit Greenland, and that Asian dust contaminates Alpine ice.We highlight that continuous geochemical and mineralogical records of paleodust provide independent and complementary approaches to reconstruct dust sources and atmospheric circulation changes.Trace/crustal elements abundance (e.g.Ti, Ba, La), mineral composition (e.g.clay mineralogy), and isotopic analysis ( 87 Sr/ 86 Sr) shows 20177 Introduction

Conclusions References
Tables Figures

Back Close
Full  Full  Full  Full core studies document enhanced atmospheric crustal dust transport during glacial/interglacial colder climates, but dust archived in Alpine, Himalayan, and Antarc-Discussion Paper | Discussion Paper | Discussion Paper | demonstrated by combining Sr and Nd isotopic fingerprinting and air-mass back-trajectory that the geochemical signature of Saharan dust transported to the Alps could be used for paleodust transport studies.Moreover, the Southern Alps act as a barrier to the transport of the southwesterly dust laden winds from the Sahara during the spring and summer seasons.Hence, the Colle Gnifetti glacier saddle, where the oldest glaciological record Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | temperature of 1480 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | , illustrating the control of the dust laden winds (PM 10 ≥10 µg/m 3 ) by the Westerlies, except for one air-mass trajectory which originated from Northeast (13 May 2009).Generally, windborne Saharan dust was directly transported by the southwesterlies winds across the Mediterranean Sea towards the Alps, although some trajectories presented longer transport pathway (e.g. 25 June 2008).Most of the dust-laden winds had origins over Algeria and to a larger extent the northcentral to north-western part of the Saharan desert (i.e.Morocco, Tunisia, Libya, and Mali, Fig. 1).Peculiar air masses trajectories reaching the Alps from the North occurred on the 12 and 14 October 2009, with 315-h back-trajectories transport pathways from North America passing over Greenland and towards the Alps from northerly directions.Although the two corresponding samples are extremely depleted in insoluble material (PM 10 <2 µg/m 3 ), their elemental crustal composition does not differ significantly from Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | that North Africa was the most important supplier of dust to mid-latitudes in the Northern Hemisphere.Long-term and continuous records of airborne dust characteristics in climatic archives provide new perspectives on our understanding of seasonal atmospheric circulation changes and large-scale atmospheric teleconnection patterns (e.gDiscussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Figure 1.

Fig. 4 .
Fig. 4. X-ray diffraction (XRD) spectra of a Jungfraujoch sample (JFJ, upper spectrum), and from two Colle Gnifetti ice core (CG) samples enriched in Saharan dust.The minerals and the Miller indices associated with the major peaks are indicated.P: paragonite, and Pk: palygorskite.
• C in static mode using the virtual amplifier design to cancel out biases in gain calibration among amplifiers. 87Sr/ 86 Sr values were internally corrected for fractionation using a 88 Sr/ 86 Sr value of 8.375209.Raw values were further corrected for external fractionation by a value of +0.03‰, determined by repeated measurements of the SRM987 standard ( 87 Sr/ 86 Sr=0.710250).External reproducibility (1 s) of the SRM987 standard is 7 ppm.Nd was loaded on double Re filaments with 1M HNO 3 and measured in static mode with the virtual amplifier design.Nd value of 0.7219 and the 144 Sm interference on 144 Nd was monitored on the mass 147 Sm and corrected by using a 144 Sm/ 147 Sm value of 0.206700.External reproducibility (1σ) of the JNdi-1 standard (Tanaka et al., 2000) is <5 ppm. 144

Table 1 .
Sample name, age of sample, 87 Sr/ 86 Sr and 143 Nd/ 144 Nd isotopic compositions with 2σ (2 standard errors of the mean), and Nd isotopic ratios expressed as epilon units.

Table 2 .
X-ray mineral compositions for the Colle Gnifetti (CG) ice core samples (upper part), and for the analog samples collected at the Jungfraujoch (JFJ) (lower part).K/C for kaolinite/Chlorite ratio.The symbol plus reflects the abundance of minerals.

Table 3 .
The backward trajectory arrival date and source countries of the Sahara dust events sampled at the Jungfraujoch (JFJ), PM 10 values, and mean grain-size.