**6. Conclusion**

*Glaciers and the Polar Environment*

or − 0.4% y<sup>−</sup><sup>1</sup>

which was here included for consistency with the 1975 and 1991 outlines drawn on the maps. The rate of change of Lys Glacier found in this study (−0.44 kmy<sup>−</sup><sup>1</sup>

in the other alpine countries [14, 46, 47]. The increase in debris cover is lower than that reported by Azzoni et al. [9] for the Ortles-Cevedale region, where 38 glaciers were reported to have on average a 13.3% higher proportion of their area covered in debris in 2012 (reaching 30%) than 2003, while Lys Glacier went from 7.9% in 1988 to 12.4% in 2000, decreasing again to 8.5% in 2014. These values are also lower than those reported by Shukla et al. [5] for Samudratapu Glacier in Indian Himalayas, where debris cover nearly doubled over less than 3 years and in line with observations of glaciers in Caucasus by Stokes et al. [11], who describe a 3–6% increase in debris cover and a 57% increase in supraglacial and proglacial lake area. The formation of lakes on stagnating debris-covered glacier tongues was also observed by Kirkbride and Warren [48] and is attributed to the presence of ice-cored moraine which prevents meltwater runoff and favors the accumulation of water in depressions left by melting ice. Similarly to Stokes et al. [11], we also found that debris cover has not halted glacier retreat, counter to the evidence that a thick debris cover is known to reduce ablation (see, e.g. [49]). A field campaign conducted in 2006 revealed that debris thickness is generally above 10 cm and up to 60 cm, well above the critical threshold for which the insulating effect prevails on the albedo effect [49]. Thus, while it is possible that mass wasting would have been even higher without debris, it is more likely that glacier retreat occurred owing to the presence of ice-contact lakes and cavities, which enhance melt through backwasting [50]. Unlike retreat rates, mean thickness and volume changes for Lys Glacier are noticeably lower than in similar studies conducted on alpine glaciers: D'Agata et al. [21] report a decrease in ice thickness of −14.91 m for glaciers in Sondrio Province,

Central Italian Alps, from 1981 to 2007, corresponding to −0.57 my<sup>−</sup><sup>1</sup>

the Swiss Alps, Fischer et al. [37] report an area-weighted mass balance of −0.62 m

analyzed in the study by Fischer et al. [37], although a few Swiss glaciers do share a similar geodetic mass balance. The geodetic mass balance of Lys Glacier would also be lower than that reported by Berthier et al. [51], i.e. –1.05 ± 0.37 m w.e. y<sup>−</sup><sup>1</sup>

glaciers in the Mont Blanc massif using ASTER, SPOT and Pleiades DEMs between 2000 and 2014. If we exclude the possibility that debris cover has slowed down thinning for reasons stated above, the large differences and the apparent low average elevation changes of Lys Glacier can be explained by (1) the relatively large size of the glacier accumulation basins and the presence of seasonal snow on the Pleiades image (**Figure 2a**) and (2) interpolation errors, especially in the oldest cartographic DEM, also seen particularly affecting the accumulation basins of the glacier where little change is expected to occur over the years [38], while we observe areas with apparent thickening of 20 m and up to 75 m. Considering the glacier tongue, the glacier thinning is however evident: to compare our findings against those of Rota et al. [27], we selected an area below 2660 m a.s.l., corresponding to 0.7 km2

The average ice loss was computed as 62.92 ± 0.81 m from 1991 to 2014, equal to a

suggesting an increase in the glacier tongue thinning rates. Our values for the glacier tongue are also in line with the findings of Mölg and Bolch [52] who reported an

between 1925 and 1953 and − 0.42 my<sup>−</sup><sup>1</sup>

 from 1980 to 2010, while the geodetic mass balance of Lys Glacier (using a conversion factor of 0.85 accounting for the average density of ice and firn as done

estimated thinning of Lys Glacier to be 0.19 my<sup>−</sup><sup>1</sup>

by [37]) would be −0.16 m w.e. y<sup>−</sup><sup>1</sup>

thinning rate of 2.74 ± 0.03 my<sup>−</sup><sup>1</sup>

average elevation change of −67 ± 5.3 my<sup>−</sup><sup>1</sup>

for a larger area and a longer period, between 1946 and 2005.

[27], i.e. -1.32 my<sup>−</sup><sup>1</sup>

, while we

for

.

between 1953 and 1994,

. In their study of all glaciers in

, which is at the low end of the scale for glaciers

, and higher than the values reported by Rota et al.

for Zmuttgletscher (Swiss Alps), albeit

) is however comparable to that of alpine glaciers, both in Italy and

**182**

w.e. y<sup>−</sup><sup>1</sup>

In this study, we analyzed the evolution of Lys Glacier, one of the largest glaciers in the Italian Alps, by looking at a variety of parameters: terminus fluctuations were studied from historical sources and glaciological bulletins from 1812 to 2017; changes in surface, debris cover and area of supraglacial/proglacial lakes together with volume changes were examined from cartographic and remote sensing datasets from 1975 to 2014. The glacier length variations were found to be similar to those of large glaciers in the Alps such as Mer de Glace and Unterer Grindelwald, indicating a similar climatic setting in spite of the distance of these glaciers; the worst conditions for the glacier development occurred after the end of the Little Ice Age and since 1985 (−443 m from 1985 to 2017) reflecting increasing temperatures as seen from the closest weather station located at Gressoney d'Ejola. Overall, Lys Glacier has retreated by almost 1.6 km since the LIA.

All the other glaciological findings point to a strong glacier reduction, which is interpreted as an evident impact of climate change: the rate of area change was −0.04 km2 y<sup>−</sup><sup>1</sup> since 1988, while glacier volume decreased by −47 × 106 m3 from 1991 to 2014. The glacier debris cover increased from 1988 to 2000, when it covered 12.4% of the glacier area, and then started decreasing again, as a result of glacier shrinking, while the area of the proglacial lakes grew exponentially over the same period. We consider the changes in area and debris cover as highly reliable in view of our accuracy assessment (max 2% error in the glacier outlines and accuracy between 90% and 100% when mapping debris cover), while the uncertainty in volume variations is larger because of the lower quality of the input DEM from 1991.

In view of the present conditions of the glacier, which prevent reaching the glacier tongue, remote sensing remains as the only viable option to investigate the glacier variations in the future, while the detachment of the glacier tongue has further complicated studying terminus fluctuations. To further our understanding of the glacier past conditions, other historical sources should be considered, including pictorial documents to lengthen the record of glacier terminus position and aerial photography from the past century to provide more accurate estimates of volume changes.
