Effects of topographic and meteorological parameters 2 on the surface area loss of ice aprons in the Mont-Blanc massif (European Alps)

15 Ice aprons (IAs) are part of the critical components of the Alpine cryosphere. As a result of the 16 changing climate over the past few decades, deglaciation has resulted in a surface decrease of IAs, 17 which has not yet been documented, except for a few specific examples. In this study, we quantify 18 the effects of climate change on IAs since the mid-20 th century in the Mont-Blanc massif (western 19 European Alps). We then evaluate the role of meteorological parameters and the local topography 20 in the behaviour of IAs. We precisely mapped the surface areas of 200 IAs using high-resolution 21 aerial and satellite photographs from 1952, 2001, 2012 and 2019. From the latter inventory, the 22 surface area of the present individual IAs ranges from 0.001 to 0.04 km 2 . IAs have lost their surface 23 area over the past 70 years, with an alarming increase since the early 2000s. The total area, from 24 7.93 km 2 in 1952, was reduced to 5.91 km 2 in 2001 (-25.5 %) before collapsing to 4.21 km 2 in 25 2019 (-47 % since 1952). We performed a regression analysis using temperature and precipitation 26 proxies


39
The predicted shift in climate dynamics over the next decades will undoubtedly have severe  However, many of these variations result from different morphometric (size, shape, length) and 55 topographic (altitude, slope, aspect, curvature, terrain ruggedness) characteristics. 56 Several studies have been devoted to understanding the linkage between topographic factors and 57 the response of glacier/ice bodies (e.g., Davies Guillet and Ravanel (2020) showed that IAs in the MBM have lost mass since the Little Ice Age 90 (LIA). Based on six different IAs, their study also showed an acceleration in the shrinkage since 91 the 1990s. They linked the loss of IA area with meteorological parameters, mainly air temperature 92 and precipitation. It was thus the first documented evidence that IAs have been losing ice volume 93 due to the changing climate. However, since this study was local and based on only a few IAs, the 94 authors could not consider other factors, such as the local topography critical for small glacier 95 bodies (Hock, 2003;Laha et al., 2017). 96 Thus, to overcome these limitations, we propose a large-scale analysis to ascertain the relationship   increased, along with an increase in bare ice areas. In some instances, ice volume loss leads to 150 instability of steep slopes, and serac falls from the front of warm and cold glaciers are more 151 frequent (Fischer et al., 2006). This latter process can be typical during the warmest periods of the 152 year (Deline et al., 2012). Warming trends also intensify moraine erosion, resulting in an increase 153 in rockfall and landslide events (Deline et al., 2015;Ravanel et al., 2018). Degradation/warming 154 is another critical concern for permafrost (e.g., Haeberli and Gruber, 2009). This section describes all the datasets obtained from diverse sources used in this study ( Since one of the main aims of our study was to perform a joint analysis of the behaviour of small 164 ice bodies and the local topography, it was paramount to have a robust high resolution and accurate

172
The second part of the processing involved accurately co-registering the newly built DEM with an 173 existing reference DEM of high precision and accuracy. For this purpose, we used the automatic 174 DEM co-registration methodology given by Nuth and Kääb (2011). As a 'reference', we used a 2 175 m LiDAR DEM for the area around the Argentière glacier (8 * 2.5 km spatial extent) (Fig. 3a) 176 built by the Institut des Géosciences de l'Environnement (IGE) to co-register the 'source' 4 m 177 Pleaides DEM (Fig. 3b) generated in the previous step. A precisely co-registered, high-resolution, 178 robust 4 m DEM was obtained at the end of the processing steps. More detailed information about 179 the processing parameters for DEM generation and co-registration can be found in Kaushik et al. 180 (2021). We used this DEM to compute topographic parameters like slope, aspect, curvature, 181 elevation, TRI, mean annual rock surface temperature (MARST) and direct solar radiation. This study relies on high-resolution aerial and satellite images (Table 1). Working with data from 186 different sources allowed us to tap into the wealth of data for comparison. Spanning over seven   The first part of our analysis follows a similar methodology followed by Guillet and Ravanel 228 (2020) in their analysis. Like their study, we used homogenized weather records from the Col du where α = 0.87 (slope) , β = -7.7 o C (intercept) and r (residuals) with zero mean.

244
No transformation for the precipitation values was performed as this relation is tough to establish 245 and not always linear (Smith, 2008 comparison, it was essential to interpolate the missing data for the six years before 1958 (Fig. 4).

260
Like the previous methodology, we looked for a linear relationship between the SAFRAN 261 temperature data (at 2400 m a.s.l. elevation belt) and the GSB temperature data. We again found 262 a strong correlation between the two datasets ( Fig. 5) which helped us transform the data using:  DSRθ,α = SConst * (β m(θ) ) * SunDurθ,α * SunGapθ,α * cos(AngInθ,α), for solar radiation indicate higher insolation, while lower values suggest low insolation. We prefer 336 DSR over the aspect for our analysis to avoid bias due to local shading on sun-exposed faces, 337 considering the slope angle associated with the aspect. representing a steady state.

380
The values for MARST are calculated in o C and, for our study region, range from -12 to 10 o C.

381
MARST is also an important criterion to check for the very likely presence of permafrost below 382 the IAs, which likely allows the formation and existence of IAs.  to 1) in the present context indicates that the relative surface area loss of IAs between the two 566 periods is comparatively less than that of IAs whose ratio is closer to 0. A value larger than 1 567 indicates an increase in the surface area over time.

568
From the results, we do not see a strong correlation (r = 0.73) between the modelled area (from 569 GSB transformed climate data) and the measured area for the 200 IAs spread across the MBM 570 (Fig. 14 a). However, the correlation improves significantly (r = 0.86) when we use the SAFRAN 571 data based on different elevations and remodel the surface area for each IA (Fig. 14 b). This can 572 be seen from the values of R 2 , Pearson's r, RMSE and the p-value estimates from the T-test 573 achieved from both datasets ( Table 2) Table 2. factors. It indicates that the effect of rock surface temperatures on the area loss of IAs is not strong 609 on a regional scale. (Fig. 15c; Table 3). However, this relationship needs to be examined in a more 610 site-specific and localized area to understand better its impact on the surface area loss of IAs. We 611 also observed that the correlation was higher for a more extensive observation period  612 than for shorter periods. This could suggest that the influence of rock surface temperatures 613 potentially becomes more prominent with a more extensive observation period.

614
A similar analysis of IAs area loss with the TRI showed a weak positive correlation ( Fig. 15d; 615 Table 3). An increase in TRI values (i.e. increase in terrain ruggedness) may result in more ice 616 area loss on a site-specific scale, but this relationship is hard to observe globally. Like the results 617 from the analysis with MARST, the strongest correlation was again observed for the largest study 618 period. Further, like the TRI, we also found a weak correlation between the terrain slope and 619 curvature with the surface area loss of IAs. We must note that our criteria for selecting IAs already 620 limit us to areas with slope angles steeper than 40 o ( Fig.15e; Table 3). Hence it was difficult to 621 observe any significant impact of terrain slope on the rate of area loss of IAs. Similarly, terrain 622 curvature seems to have the most negligible impact ( Fig. 15f; Table 3). As cited in Sect. 4.2.

623
previous studies may have shown that terrain curvatures could play an essential role in the erosion 624 and accumulation dynamics on steep slopes, but this is not the case for IAs in the MBM. We 625 performed the last comparison between the relative surface area loss of IAs with their initial area.

626
Our results were similar to the one Lopez et al. (2010) reported, as we did not find any correlation 627 between the two quantities ( Fig. 15g; Table 3).

734
Hence IAs do not directly participate in feeding the larger glacier systems below them. However, 735 the avalanches triggered above can bring fresh snow/debris and lead to erosion or deposition on 736 the IA surface. We expect this factor to also play a role in the area change dynamics of the IA,