Dating of ice cores from temperate non-polar glaciers is challenging and often problematic. However, a proper timescale is essential for a correct interpretation of the proxies measured in the cores. Here, we introduce a new method developed to obtain a sub-seasonal timescale relying on statistically measured similarities between pollen spectra obtained from core samples and daily airborne pollen monitoring samples collected in the same area. This approach was developed on a 10 m core retrieved from the temperate-firn portion of Alto dell'Ortles glacier (Eastern Italian Alps), for which a 5-year annual/seasonal timescale already exists. The aim was to considerably improve this timescale, reaching the highest possible temporal resolution and testing the efficiency and limits of pollen as a chronological tool. A test of the new timescale was performed by comparing our results to the output (date of layer formation) of the mass balance model EISModel, during the period encompassed by the timescale. The correspondence of the results supports the new sub-seasonal timescale based on pollen analysis. This comparison also allows us to draw important conclusions on the post-depositional effects of meltwater percolation on the pollen content of the firn core as well as on the climatic interpretation of the pollen signal.
Ice core dating is crucial for the interpretation of the paleoclimatic and paleoenvironmental proxies contained in glacial archives. When combined with the detection of absolute temporal horizons, annual layer counting is the most accurate technique to date ice cores (Thompson et al., 2013). However, low snow accumulation and/or post-deposition effects (e.g., meltwater percolation) hamper the detection of annual layers, especially in temperate glaciers where annual signals are most often smoothed (Eichler et al., 2001). In addition, no proxies had so far a level of temporal precision that allows studying past seasonal changes in detail, while a sub-seasonal temporal resolution would be desirable to reconstruct these changes in detail.
Several studies conducted on glaciers worldwide have proven that pollen
(Nakazawa et al., 2004, 2005, 2011, 2015; Santibañez et al., 2008;
Uetake et al., 2006) and stable isotopes (Gabrielli et al., 2008; Haeberli
et al., 1983; Thompson, 1980; Dansgaard, 1964; Vareschi, 1934) are valuable
proxies to detect seasonality in ice cores. However, annual layer counting
from oxygen and hydrogen isotopes ratios is a far more common chronological
tool than the palynological approach, the potential of which remains largely
unexplored. The reason for the scares use of cryopalynology has functional
and conceptual basis. The main issue is that pollen analyses requires a
minimum ice sample size of up to 1 L (e.g., Burogois, 2000), which is
problematic to obtain because sample volume from ice cores is very limited,
especially when working at high resolution. However, studies from the Altai
(Nakazawa et al., 2015, 2011, 2005, 2004) and the European Alps (Festi et
al., 2015; Bortenschlager, 1970a, b; Vareschi et al., 1934) suggest that
this is still a limit for clean samples obtained from polar ice caps,
whereas in low and midlatitude glaciers, the minimum sample size can be
reduced to 10–30 mL (Festi et al., 2015; Nakazawa et al., 2011, 2005)
thanks to the proximity of the source vegetation and the consequent much
greater pollen deposition. An additional limit is that pollen analyses are
time consuming and work intensive because they imply manual identification
and quantification of pollen grains. In order to overcome this latter issue
Nakazawa et al. (2011, 2005, 2004) adopted a simplified approach,
focusing on three main taxa (Pinaceae, Betulaceae and
In Festi et al. (2015) we developed an efficient method to detect seasonality employing a 10 m shallow core extracted in June 2009 from the Alto dell'Ortles glacier (South Tyrol, Italy) (Fig. 1) and we improved the existing timescale based on the isotopic composition of the core (Gabrielli et al., 2010). In Festi et al. (2015) we conducted accurate taxonomical identification and implemented a statistical approach consisting in performing a principal component analysis (PCA) on pollen concentration values and extracting the three principal components (PCs) indicative of the three flowering seasons. As each PC summarizes the seasonal information of the pollen assemblage, score values of PCs indicate seasonal/annual patterns and enable the identification of seasonal and annual firn layers (Fig. 2a). In Festi et al. (2015) we also discussed that the main pollen input on the glacier likely comes from the near valleys, as the pollen spectra from the ice samples and from valley floor at Solda's monitoring station (Fig. 1) are very similar.
The present study uses the palynological data discussed in Festi et al. (2015) to develop a refined and innovative new pollen-based method to date ice core samples at a sub-seasonal resolution. The aim was to improve the chronology enhancing resolution, dating efficiency and coherence of seasonal patterns (i.e., overlapping of components).
Combined with mass balance modeling, the new high-resolution results also provide new insights into the processes controlling the formation and preservation of the palynological signal in firn and ice.
The Alto dell'Ortles is the highest glacier of South Tyrol (Italy) in the
Eastern European Alps (46
In this study, we also use air temperature data recorded by a standalone data logger placed on the Mt. Ortles at 3835 m a.s.l., as well as meteorological (air temperature and precipitation) and airborne pollen data collected at the meteorological station of Solda (Festi et al., 2015). The station is located 4.5 km northeast of the Mt. Ortles at an altitude of 1850 m a.s.l. (Fig. 1).
This section describes (i) the novel pollen-based method “from depth-to-day”, developed to obtain a high-resolution timescale for the 2009 Mt. Ortles shallow firn core, and (ii) the approach used to obtain a core layer dating by using a mass balance model (EISModel), which serves as a comparison for the newly developed palynological timescale.
Details on the composition of the Mt. Ortles and Solda's pollen assemblages
are reported in Festi et al. (2015). Here, we use these data to develop an
enhanced resolved chronology at sub-seasonal timescale. The extensive pollen
analyses carried out on the 103 samples obtained from the 10 m core provide
a high diversity of 64 pollen types, including the main pollen types
characterizing the forest vegetation in the region (e.g.,
Solda's airborne pollen samples are characterized by their specific pollen
content on a specific day of the year (DOY), while each of the 10 cm
sequential Mt. Ortles samples is characterized by its pollen assemblage at a
specific depth in the firn core. The 3-year palynological dataset of the
Solda's aerobiological station has been considered as a representative
calibration dataset to define the flowering DOY for in the entire period
covered by the 10 m Ortles firn core (2005 to 2009). For every 10 cm
sequential Mt. Ortles sample, the three Solda's airborne samples with the
highest similarity (one per year of the 3-year Solda dataset) were
selected, using the Jaccard similarity index (Jaccard, 1901). This index was
chosen because it is asymmetrical and hence avoids the double zero problem;
i.e., it excludes similarity in case of absence of a pollen type from pairs
of compared samples. The Jaccard similarity index was calculated with the
SPSS software obtaining a matrix of similarity indices. Indices typically
presented values scaled from zero to one: the higher the value, the greater
the similarity between two samples. The lower boundary for the Jaccard index
was set at 0.5 to ensure high similarity and avoid possible mismatches.
Three potential DOYs were obtained for every Mt. Ortles sample. For each
sample, the average DOY and uncertainty (one sigma) have been calculated
(Table 1, Fig. 2b). Mt. Ortles samples with pollen concentrations
reflecting “winter” season (< 0.5 grains mL
Results of the depth-to-day matching obtained by similarity analyses between Mt. Ortles snow samples and pollen monitoring data from Solda from the years 2008 to 2010. DOY is day of the year. Samples are ordered according to their depth in the core.
In summary, by coupling the Mt. Ortles firn samples with the most statistically similar assemblage of the Solda's airborne samples, we establish a link between pollen deposition at a specific sample depth on Alto dell'Ortles and a specific DOY. This “space-for-time” substitution (depth-to-day) by pollen is the key concept of the new dating technique developed. This method is based on the assumption that there is no time lag between the flowering in Solda and the pollen deposition on the glacier thanks to the efficient uplift of pollen grains by thermic wind (Barry and Chorley, 2009). Depth values were converted to water equivalent values to enable comparison with EISModel results, using a polynomial function fitted to the 2009 firn core density sampled at 10 cm depth intervals. Samples are ordered (Table 1) according to their increasing depths from the top of the core and increasing w.e. from the bottom of the core.
Here we briefly describe the essential parts of the mass balance model used in this study (Carturan et al., 2012a). The cumulated mass balance from 2005 to 2009 at the coring site was calculated using EISModel (Carturan et al., 2012a; Cazorzi and Dalla Fontana, 1996). Before being included in the model, the raw meteorological data recorded at Mt. Ortles and Solda were checked and validated against other meteorological weather stations located in the proximity of Mt. Ortles (Madriccio at 2825 m and Cima Beltovo at 3328 m). The precipitation data recorded at Solda were corrected for gauge undercatch errors, using the procedure described in Carturan et al. (2012b).
The mass balance model simulates accumulation and melt processes at hourly
time steps. Snow accumulation was calculated from the precipitation data
recorded at Solda, extrapolated to the elevation of the study site using a
precipitation linear increase factor (PLIF; % km
Ablation was calculated by means of an enhanced temperature-index approach,
using the clear-sky shortwave radiation computed from a digital elevation
model (light detection and ranging, or lidar, survey of 2005, provided by the province of Bolzano) as a
distributed morpho-energetic index (Carturan et al., 2012a). The melt rate
MLT
For each snow layer deposited (i.e., the water equivalent that accumulates at the surface of the snowpack during an hourly time step), the model provides its time and date of formation as well as the air temperature during its deposition.
Results of depth-to-day match of firn and Solda's samples are shown in Table 1 and Fig. 2b. Each date illustrates the time period encompassed by the sample, with the amplitude of the error bars indicating the number of days potentially included in each sample. The surface sample was dated to 6 June 2009, in good agreement with the fact that the core was extracted in the first half of June. The dates cluster in five groups representing five vegetation periods and are distributed in chronological order within each year. Few inversions are present (i.e., samples at 78 cm w.e. in 2006; 175 cm w.e. in 2007; 324 cm w.e. in 2008; 434, 448 and 452 cm in 2009), but the estimated dates are generally within the uncertainty of adjacent samples. Such inversions are reported in Table 1, as they potentially represent disturbances in the snow deposition (see Sect. 5.2). We observe a substantial difference in the pollen distribution patterns in the snow, both within and among flowering seasons. Flowering seasons stand out distinctively as layers with high pollen concentration, variable interannual thickness and seasonality patterns. According to their thickness and vertical distribution of pollen, the flowering years cluster in two groups: 2005 and 2006 vs. 2007, 2008 and 2009. To be specific, the flowering seasons of 2007 and 2008 correspond to very thick firn layers (76 and 65 cm w.e. respectively), into which pollen is distributed with a clear seasonal pattern. The flowering seasons of 2005 and 2006 are characterized by (i) a significantly lower firn thickness (12 and 34 cm w.e. respectively), (ii) the occurrence of a thin lower layer with a distinct spring pollen content and (iii) a thin upper layer containing mixed spring/summer pollen. In contrast, the non-flowering seasons (October to February) are clearly visible as firn layers which are free (or nearly free) of pollen and present significant differences in thickness. Winter 2007/2008 is the thinnest, with only 33 cm w.e., followed by 2006/2007 with 44 cm w.e., 2005/2006 with 60 cm w.e. and finally 2008/2009 with 91 cm w.e.
Mass balance modeled by EISModel compared with
The cumulated mass balance simulated by the model shows high variability of accumulation and melt rates during the period from 2005 to 2009 (Fig. 3). In particular, there is a strong difference between the first 2 years (2005 and 2006), during which snow accumulation mostly occurred in late summer after weeks with high ablation, and the last 3 years (from 2007 to 2009), characterized by higher accumulation rates and a much lower summer ablation. In 2005, the ablation removed snow layers accumulated between 9 April and 31 July. The same happened in 2006 to the snow layers accumulated between the end of March and the end of July.
Comparison between cumulated mass balance modeled by EISModel and obtained by pollen dating. Alternating grey/white bars have a 6-month duration and roughly indicate the warm/cold season. Horizontal dash lines indicate melt layers.
Given that the precipitation regime in the Mt. Ortles area is characterized by a winter minimum and a summer maximum (Schwarb, 2000), corresponding maxima and minima in the snow accumulation rate are expected at the study site, in absence of significant ablation. This behavior was indeed modeled in the hydrological years 2006–2007 and 2007–2008. In 2008–2009 the modeled accumulation rate was high also during winter as a result of abundant winter precipitation in the Italian Alps.
The depth-to-day method and the PC approach (Festi et al., 2015) provide converging evidences that the core encompasses five vegetation periods that can be assigned to specific years by counting back from the core's extraction date in June 2009 (Fig. 2a and b). However, the new method shows clear improvements at various levels: (i) it provides a sub-seasonal time-resolved dating; (ii) it estimates the timespan encompassed by each dated sample (represented by the amplitude of the error); and (iii) it resolves the main issue of the PC method, namely extracting a coherent chronological succession when PCs overlap, e.g., in years 2005 and 2006. For example, in the year 2005 the overlapping of PCs makes the interpretation of the seasonality problematic. In contrast, the depth-to-day method provides three distinctive dates in chronological order, encompassing a time period from April to July (Fig. 2b). A similar situation is found in the year 2006, which mainly corresponds to an overlapping of PCs between 80 and 85 cm w.e. of depth. This overlapping follows a very low rise in the spring component and precedes an equally low increase in the late summer component. The depth-to-day method provides a better chronological succession of the samples by dating the first sample of 2006 to March, a second cluster of three samples to May–beginning of June (corresponding to the PCs overlapping) and the final sample to the middle–end of June. The central cluster of dates also includes an inversion, likely indicating disturbances in the pollen and/or in the snow deposition, possibly caused by melting or wind erosion and redisposition (see also Sect. 5.3). Also in 2007 the succession of dates provides a more straightforward timescale than the components do, solving the case of overlapping components between 160 and 180 cm w.e. of depth. The ability of the depth-to-day method to produce a detailed dating for layers otherwise characterized by the PCs overlapping suggests that this method is more efficient in detecting changes in the pollen assemblage. In general, component overlapping can be expected because the flowering of the different plants occurs in a continuum, without pauses between spring, early summer and late summer season, and therefore without sharp changes in the airborne pollen assemblage. The only break in the flowering is represented by the autumn and winter, as in this period there is no local pollen production, leading to firn layers with no, or sporadic, pollen occurrence. In addition, the PC overlapping may also result from sampling two flowering seasons in a single snow/firn sample.
In summary, the day-to-depth method significantly improved the timescale by Festi et al. (2015), providing a clear and intuitive representation of the chronological succession, showing a better efficiency in detecting changes in pollen assemblages that allows establishing a sub-seasonal-resolution timescale and providing the estimation of the number of days included in each dated sample.
Application of the depth-to-day pollen concept and mass balance
modeling provides two independent datings of the snow and firn layers of the
10 m firn core drilled in June 2009. Layer dating obtained with EISModel and
pollen analysis is in very good agreement (Pearson correlation coefficient
The two methods provide converging evidence that two distinct types of warm seasons occurred during the investigated period, i.e., summers with high ablation in 2005 and 2006 and summers with high accumulation in 2007 and 2008 (Figs. 4 and 5). The high agreement between pollen dating and EISModel results further confirms that in the Ortles region pollen grains are very efficiently transported upward from the vegetation source to the glacier with no/negligible time lag. This is due to the vicinity of the vegetation to the ice body and to the steep altitudinal gradient (Fig. 1). Such efficiency, however, cannot be assumed to be a common feature for every glacier–periglacial environment. In fact, Nakazawa et al. (2015) observed a significant delay (about 1 month) between pollen production and its deposition on a glacier in the Mongolian Altai. In this context, the application of the depth-to-day method can be problematic; however, if the time lag between pollen production and deposition is known it can be integrated in the dating model.
Comparison of the winter and summer mass balance as obtained by pollen dating and modeling (EISModel).
Comparing the detailed chronological distributions obtained by the palynological methods and by the EISModel helps in the interpretation of problematic features such as the pollen timescale inversions (empty squares in Fig. 2b). Theoretically, chronological inversions can potentially be of two origins. (a) There might be a statistical artifacts due to the limited time period (3 years) covered by the available data from the aerobiological station of Solda. A longer training period for our palynological method would probably have enabled a better dating accuracy. (b) There can also be physical processes that have altered the original deposition of pollen and snow: redistribution of surface snow layers caused by wind erosion and redisposition and mixing of pollen in melting layers. The high-altitude location of the Ortles core site favors strong wind redistribution because it is not sheltered by the surrounding relief and because the low temperature keeps the snow dry (and therefore mobilizable) for long periods, also in spring and summer. The effects from surface melt and water percolation are discussed in Sect. 5.3.
For the years 2005 and 2006 the EISModel calculations indicate the occurrence
of strong summer ablation, leading to the removal of spring and early summer
layers for a thickness of about 30 cm w.e. The occurrence of a 0.45 cm thick
ice lens at the bottom of the core and a thick 6.3 cm lens at 50 cm w.e.
(Fig. 2d) is indication of water percolation and refreezing due to summer
melting. Despite water percolation, dating of the 2005 samples provided
dates which are in chronological order, with no evidence of pollen
displacement. In the 2006 sequence, the only hint for a minor displacement
of the pollen is a cluster of three samples dated to May–June, which
presents small inversions. While it is hard to assign the origin of such
inversions to either percolating water or a limited training set for the
depth-to-day method, a minor displacement of the pollen within these layers
cannot be excluded. In contrast, a major displacement can be excluded
because there is no evidence of pollen displacement downward to the early
spring layers in both 2005 and 2006, as well as in the 2005/2006 winter
strata, despite the fact that ablation reached 30 cm w.e. in summer 2006. In
fact, pollen concentration in this winter layer remains below 0.5 pollens mL
Our results on the resilience of pollen to percolating water also agree with
observations by Nakazawa and Suzuki (2008) in a snowpack study on the
Norikura Highlands (1590 m a.s.l.) in Japan, where these authors found that
during melt phases pollen concentrates on the surface of the snow and is not
transported downward by meltwater. This outcome supports the idea that
pollen grains are too big (5–200
These divergent conclusions on the effect of percolating water on pollen
grains point to the fact that more specific studies on this phenomenon are
needed, as it is likely influenced by laboratory design conditions, the
natural interplay of local micro-climatic conditions, physical
characteristics of snow and firn, and possibly pollen grain size and shape.
For example, on the Alto dell'Ortles, fresh winter snow usually accumulates
in windy conditions and has a density of 300 kg m
The combined use of the pollen and EISModel timescale further corroborates the finding by Festi et al. (2015), which improved the dating of the 2009 shallow core obtained from chemistry-based seasonality (Gabrielli et al., 2010) that is more likely to be affected by meltwater percolation. Therefore, an approach that combines at least two of these methods turns out to be a more reliable approach for dating firn cores from temperate glaciers in the Alps. This may be valid also for cores retrieved from other ice bodies located in an environmental setting similar to the European Alps, where the vegetation is close to the glaciers and leads to abundant pollen deposition.
The comparison between pollen content of sequential firn samples and cumulative mass balance modeling (Figs. 3 and 5) led us to argue that pollen has a good potential not only for dating but also for inferring the impact of past climatic conditions in firn and ice cores at seasonal resolution, as already recognized by Nakazawa et al. (2015). Based on the pollen content in the Mt. Ortles strata, three main types of pollen assemblages can be identified and correlated with the corresponding seasonal climatic conditions inferred by EISModel. (i) Thin, pollen-rich layers have a fairly clear date order but PCs overlapping (i.e., 2005 and 2006); EISModel indicates that such layers are the result of intense summer ablation thus pointing to warm and dry summer periods. (ii) Certain thick layers have a significant pollen concentration and well-distinguished succession of dates (i.e., 2007 and 2008); for these layers EISModel shows a spring/summer snow deposition characterized by minimal melting, generated by abundant precipitation and low temperatures. (iii) Other thick layers have no (or nearly absent) pollen, representing snow deposition during the autumn and winter seasons.
Figure 5 shows the correlation between the inferred summer and winter mass balance according to the pollen dates and EISModel. The pollen-based seasonal snow accumulation has been defined for each flowering year (summer in graph) as the depth in w.e. encompassed between the first and last datable sample of each year, while winter (non-flowering season) mass balance has been calculated as the w.e. accumulated between consecutive summers (flowering seasons). For a direct comparison, the same dates for each season were used for EISModel. Figure 5 shows that the thickness of the warm/cold season layers dated by pollen analyses and modeled by EISModel gives a fairly good qualitative indication of the amount of seasonal precipitation: summer 2005 and 2006 stand out for the low accumulation in w.e. with both methods; in contrast, summers 2007 and 2008 are inferred by both methods as high accumulation summers. Similarly, among the cold seasons winter 2008–2009 stands out for its higher w.e. thickness correlated to exceptionally high precipitation.
Undoubtedly, this is only the first step towards a qualitative climatic interpretation of the pollen signal and further investigations on longer climate series and pollen sequences from glacier cores are required in order to support this evidence. Further studies are also required to detect as many combinations of pollen assemblage types and ice layers as possible and to further correlate them with the corresponding conditions of formation. In fact, there are several other possible combinations, e.g., thin layers with high pollen concentration formed during dry (but not particularly hot) summer periods or thick layers with high pollen content but inconsistent or mixed sequence of seasonal components, deriving from relatively low winter accumulation and possible blending of 2 or more years. Our results suggest that combining the classical stable isotope method with palynological analyses not only enhances the accuracy of the ice core dating but also offers the potential to provide a calibration set to obtain qualitative paleoclimatic information at seasonal resolution. Clearly, the feasibility of this approach in deeper ice cores depends on the amount of ice available for pollen analyses, the resolution achievable during sampling and the condition of no delay between pollen production and deposition. When applying this qualitative method to deeper ice cores a thinning of the layers by compression must also be taken into account while comparing ice or firn accumulated in different epochs. Finally, the feasibility of the approach in deep ice cores also relies on potential regional shifts in pollen speciation.
In this study, we have proposed a sub-seasonal timescale for the 10 m Mt. Ortles firn core. The day-to-depth method significantly improved the
timescale by Festi et al. (2015), providing a clear and intuitive
representation of the chronological succession, proving a better efficiency
in detecting changes in pollen assemblages that allows establishing a
sub-seasonal-resolution timescale, and providing the estimation of the
number of days included in each dated sample. The method can be applied to
all types of glaciers, regardless of their thermic state, provided the
proximity of the pollen source, with negligible source–sink lag, the
existence of flowering seasons (typical of midlatitudes) or a clear
contrast between a flowering and non-flowering season (typical of the
tropics and Equator due to the alternation of dry and humid season), and the
support of modern pollen monitoring data. We also show that a 3-year
training set of pollen monitoring is sufficient to provide meaningful
comparison with glacier samples. This consideration becomes particularly
relevant when working in glaciated regions that are not covered by the
pollen allergy network, which is the main source of modern pollen data, e.g.,
World Allergy Organisation (WAO;
The combined use of a mass balance model and pollen-based dating methodology brings compelling evidence that on the Alto dell'Ortles glacier pollen grains are resilient to downward transport by percolating meltwater, even in the case of strong melting as in 2005 and 2006. The independent dating of firn layers by mass balance modeling and pollen match well, and highlight detectable intra-seasonal and interannual differences of high-altitude snow accumulation rates on Mt. Ortles. More specifically, we found evidence of peculiar types of pollen distribution in firn layers, that may be related to well defined weather conditions (e.g., warm-dry, warm-humid or cold-humid weather). These results reveal the good potential of pollen for inferring past climatic conditions at a sub-seasonal resolution in ice core records. Future studies could focus on joined fieldwork observations of meteorological variables (temperature, snow, wind, etc.) with measurements of snow accumulation/ablation, pollen accumulation, signature of isotopes and other chemical species as well as on monitoring of post-depositional effects (snow redistribution, water percolation, sublimation) on the proxies in order to directly test the assumptions made in the present study. This type of approach opens the possibility of gaining new insights in this paleoclimatic archive under “borderline” environmental conditions for proxies reliability.
Palynological data of the Ortles firn core can be found at
D. Festi, W. Kofler and K. Oeggl performed palynological analyses on the Ortles samples and developed the pollen-based timescale. E. Bucher performed the pollen analyses at the aerobiological station of Solda. L. Carturan, F. Cazorzi and F. de Blasi processed the raw meteorological data and mass balance measurements and carried out the mass balance simulations using EISModel. P. Gabrielli planned the logistic of the field campaign, drilled the firn core and coordinated the processing of the samples. D. Festi, L. Carturan, P. Gabrielli and K. Oeggl prepared the manuscript with contributions from all co-authors.
The authors declare that they have no conflict of interest.
We thank the Autonome Provinz Bozen Südtirol, Abteilung Bildungsförderung, Universität und Forschung, for financial support to the project PAMOGIS (Pollen Analyses of the Mt. Ortles Ice Samples). This work is a contribution to the “Ortles project” – a program supported by NSF award nos. 1060115 and 1461422 to Ohio State University, the Fire Protection and Civil Division of the autonomous province of Bolzano in collaboration with the Forest Division of the autonomous province of Bolzano and the National Park of Stelvio. This is Ortles project publication no. 8. We acknowledge the collaboration in the various phases of the 2009 and following years field operations of Hanspeter Staffler, Michela Munari and Roberto Dinale (Fire Protection and Civil Division of the autonomous province of Bolzano), Ludwig Noessig, Elmar Wolfsgruber and Claudio Carraro (Ufficio geologia e prove materiali of the autonomous province of Bolzano), Marc Zebisch and Philipp Rastner (EURAC), Jacopo Gabrieli (IDPA CNR-Venice), Roberto Seppi (University of Pavia), Thomas Zanoner (University of Padova), Karl Krainer (University of Innsbruck), Paul Vallelonga (University of Copenhagen), Michele Lanzinger, Matteo Cattadori and Roberto Filippi (Museo Tridentino di Scienze Naturali), and Silvia Forti (Istituto di Cultura le Marcelline). We thank Michele Bertò for assistance in designing Fig 2. For the logistic support, we are grateful to Toni Stocker (Alpine guides of Solda) and the helicopter company Airway. This is also BPCRC contribution no. 1562. We thank two anonymous reviewers for their comments. We are grateful to Jean-Louis Tison for his useful remarks and suggestions that helped improve the original manuscript. Edited by: J.-L. Tison Reviewed by: two anonymous referees