In order to quantify areal changes of Aparejo Glacier, a series of archival aerial and terrestrial photography, along with declassified satellite imagery, were orthorectified and geo-referenced. All these data were complemented with current high-resolution imagery. Table 2 summaries the details of each dataset. The oldest dataset corresponds to the 1956 Hycon photogrammetric survey provided by the Chilean Military Geographical Institute (IGM). The 1967 declassified stereo Corona KH-4A satellite imagery were acquired in digital format from the US Geological Survey (USGS). Aerial photographs from 1983 (Chile-60 flight) and 1996 (GEOTEC flight) were obtained from the Chilean Air Force Aero-Photogrammetric Service (SAF). All aerial and declassified satellite data were processed with the Agisoft Metashape Professional software (version 1.8). Due to the lack of Ground Control Points measured in the field, no Digital Elevation Models (DEMs) were generated from the aerial photographs of 1956, 1983, and 1996. All datasets were projected to the Universal Transverse Mercator (UTM) 19S zone.

Once processed, the outlines for Aparejo Glacier were mapped through manual digitization by the same expert user. For the 2015 and 2020 outlines, high-resolution imagery was employed (Geoeye and Worldview) at a centimetric spatial resolution (Table 2). The datasets were obtained from the servers https://yandex.com/maps, https://www.bing.com/maps and ESRI™ Basemap. Due to data restrictions from the Yandex server, the image’s acquisition date ranges from 2012 to 2013, based on a Google Earth imagery inspection. In the case of the 2015 outline, the lateral boundaries of the glacier were updated to 2015 with the help of kinematic GNSS data and the GPR data collected in the field (see section 5.3). For the 2020 outline, a hillshade model was generated following a terrestrial LiDAR scan survey data of 2020 (see Section 4.2). All these processes were developed using ArcGIS Pro 3.2.

An additional historical dataset was generated for the 12 March 1980 field survey footage (Marangunic, 1980). Following the 2020 LiDAR, data coupled with high-resolution imagery, 4 ground control points (GCPs) where obtained. The latter were employed in the Agisoft Metashape Professional software (version 1.8) for deriving a 1980 photogrammetric model generation. Seven 1980 field photos were used for that purpose, all taken from a helicopter flight after the detachment on 12 March. The generated model allowed us to delineate the pre-event glacier outline and the post-detachment glacier remnants.

In total, we manually delineate the glacier contours for 1956, 1967, 1980, 1996, 2015, and 2020. For all mentioned datasets, the uncertainty of the manual glacier delineation process was estimated based on 1-pixel size of the digitized aerial photographs and satellite imagery (Williams et al., 1997), which was subsequently multiplied by the perimeter length (Farías-Barahona et al., 2020b).

Complementary to the analysis, high resolution (sub-meter pixel size) Google Earth Pro optical imagery was employed for geomorphological assessments regarding the Aparejo Glacier current and recent conditions. 9 images in total from the summers of 2014 to 2023 were analyzed, most of them showing cloud free conditions with scarce patches of remnant snow.

On 6 June 2015 a ground penetrating radar survey was performed on Aparejo Glacier in order to determine glacier thickness, along with a GNSS survey, which allowed to estimate the surface topography and the geolocation of the radar measurements. The topographic survey was performed using two Topcon Hiper SR receivers, one static and one mobile (rover) units. The kinematic GNSS survey provided a nominal altitudinal precision of 10 cm of the data. Replicating the followed procedure and obtained results by Farías-Barahona et al. (2020b) at El Morado Glacier, later, on 10 April 2020 a LiDAR scan of the whole glacier was obtained using a REIGL VZ-6000 long-range terrestrial laser scanning. One reflector was established on a non-glacier location at the eastern margin of Aparejo Glacier. Although the LiDAR can measure at a millimeter resolution, combined with GNSS georeferencing and optical distortions from the atmosphere, the nominal vertical, and also horizontal, resolution is estimated at 10 cm. In our LiDAR survey the point cloud density was at least 50 points/m². After this data a Digital Elevation Model (DEM) was generated with a 1 m spatial resolution using CloudCompare Open Source Software. The DEM was used to estimate the minimum and maximum elevations of Aparejo Glacier throughout the analysis for each glacier outline generated. The described process was developed using the Add Surface Information tool on ArcGIS Pro 3.2. All datasets were projected to the Universal Transverse Mercator (UTM) 19S zone considering ellipsoidal elevations.

Both datasets (Table 3; Figure 3) were compared to estimate the recent glacier elevation changes for the period 2015–2020 through the direct subtraction of elevation values using the Field Calculator Tool on ArcGIS Pro 3.2 for each elevation point obtained from the GNSS survey. No interferometric nor photogrammetric data were considered for this assessment. Because of the limited spatial extent of the 2020 TLS data, no stable off-glacier areas could be surveyed. Thus, the level of uncertainty attributed to the calculations was quantified as 1 sigma of the observed elevation changes. For the pre-and post-event glacier elevation changes, the 1980 photogrammetric model was employed for a direct comparison of the glacier margins elevations from all years with available outlines. This analysis was performed employing the measuring tool on ArcGIS Pro 3.2 applied to 13 elevation profiles perpendicular to the glacier main flow-line.

The 2020 LiDAR data was employed to reconstruct the 1980 pre-detachment glacier topography. For this purpose, 13 transverse profiles were traced between the 1980 glacier margins, considering a flat surface along each profile, and assuming an elevation coincident with the western margin, which is more clearly visible on the imagery than the eastern margin. For the longitudinal profile reconstruction (Figure 3), the center of the transverse profiles was assumed to correspond to the ascending path of the 2015 GPR and GNSS profiles.

On 6 June 2015 a GPR survey was performed on Aparejo Glacier. We used an impulse Ground Penetrating Radar¹ of a central frequency of 10 MHz, with 8 m-long ½ dipole antennae and a wavelength λ_air of 32 m in air and λ_ice of 16.8 m in ice, with a pulse repetition rate of 1 kHz. The receiver stacked 1,024 traces with 256 samples per trace. Due to the shadow effect of the transmitted pulse, near-surface reflections cannot be detected and only thickness values larger than 10 m can be measured. The nominal accuracy of the ice depth data can be estimated as λ_ice/4 equal to 4 m. The data obtained were processed using Reflex Win software (Sandmeier, 2011), version 7.0. The GPR was coupled with GNSS differential positioning, as shown in Supplementary Figure S2. A crossover analysis of the GPR and GNSS data was performed, confirming a precision of 0.1 m for both surveys, as is exposed on Supplementary Tables S1, S2. A subglacial profile was generated after the GPR data representing the glacier bed. This information was then compared with the elevation obtained from the GNSS and LiDAR survey along with the 1980 photogrammetric model.

With the aim of characterizing the stable water isotopes of the glacier ice, on 6 June 2015, 16 samples of glacier ice, frozen water from a stream in front of the glacier and stream water down valley were collected for stable isotope analysis. During the third campaign of 9 April 2016, 10 additional ice samples for isotopic analysis were collected from Aparejo Glacier, including two from Glacier 51. In total, 26 samples were collected at Aparejo Valley (Supplementary Figure S1). Analyses were performed at the Isotope Analysis Laboratory, Andrés Bello University, Viña del Mar, Chile. In order to establish the heavy versus light isotopes ratios of H and O (deuterium, ²H, and ¹⁸O, respectively), samples were filtered using 0.2 μm cellulose filters and analyzed by cavity ring-down spectroscopy (CRDS) using a Los Gatos Research (LGR) triple liquid isotope analyser (TLWIA-45EP). Results were normalized to the Vienna Standard Mean Ocean Water (V-SMOW) – Standard Light Antarctic Precipitation (SLAP) scale (reported as δ¹⁸O and δD). We will refer hereafter to the Global Meteoric Water Line (GMWL) (Rozanski et al., 1993), defined as follows: δ H 2 = 8.17 δ O 18 + 11.27 %

After normalizing the isotope values the analysis focuses on variations due to altitudinal effects (Clark and Fritz, 1997).

Two main datasets of meteorological data were analyzed to determine the climatic conditions before and after the Aparejo Glacier detachment. The first comprises the upper air temperature at 850 and 700 hPa obtained from the coastal radiosonde station Quintero (−32.78333, −71.5333, 7 m a.s.l., station WMO number 85543) for the period 1 February to 31 March 1980. The second dataset corresponds to precipitation data of 5 weather stations deployed on the western side of the Andes near the Aparejo Glacier. They are San Gabriel (−33.7819, −70.2400), San José (−33.6153, −70.3506), El Yeso (−33.6767, −70.0883), Las Melosas (−33.9044, −70.1950) and Lagunitas (−33.0800, −70.2528) stations (Table 1). Daily precipitation for February and March and the Zero Isotherm Altitude were analyzed (Figure 4A). All data were obtained from the Meteorological Direction of Chile (DMC), where they passed a quality control process. Besides this control, and because precipitation in central Chile is mostly associated with passing frontal systems and, therefore, if it rains in Santiago, it is highly probably rain in the mountains, the station time series including Quinta Normal and Tobalaba (station numbers 85,577 and 85,580, respectively) were used as spatial homologation test. All monthly data were compared among each other to confirm similar intra and interannual variability. Following the latter, the annual precipitation was estimated at each station for the periods indicated on Table 1 according to an exponential adjustment defined as follows (considering a mean squared error of 1.0): Annual Precipitation mm w . eq . = 254.88 e 0.0005 H m a . s . l .

Observations performed during the first field trip on 28 May 2015 revealed the existence of continuous glacier ice outcrops at an elevation of 3,285 m a.s.l., close to the location of the former Aparejo Glacier front. Most of the observed ice had a thin surface debris cover, less than 10 cm thick (see Supplementary Figures S2A,B). During the second field trip performed on 9 April 2016, we confirmed that the observed lowermost debris-covered ice during the first field trip, coincided with the 1956 glacier front. The glacier was discovered on the former glacier basin, largely covered (more than 75% of its surface) by a thin debris cover less than 5 cm thick (Supplementary Figure S2C). Many tensional crevasses were found close to the upper glacier section, with several shear fractures at its margins, some wider than 1 m (Supplementary Figure S2C).

Figure 5 and Table 4 summarizes our results after the successive delineation of Aparejo Glacier from 1956 to 2020. Only the 1983 aerial imagery was not useful because of the presence of snow covering most of the lower two thirds of the glacier. Aparejo Glacier area varied from 0.13 ± 0.002 km² in 1956 to 0.10 ± 0.002 km² in 2020, with a probable maximum extent prior to the glacier detachment in 1980 with an area of 0.21 ± 0.002 km². At the same time, the 1980 glacier outline poses the widest width, with an average value of 220 m for the detached portion. In contrast, the lower two thirds of the glacier before and after its detachment had an average width below 100 m. In almost all years the length of the glacier was estimated to be around 1 km. Only the 1967 outline has a shorter length of 810 m. The minimum/maximum elevations of the glacier are estimated to be 3359/3790 m a.s.l. in 1980 and 3350/3759 m a.s.l. in 2020.

The post-event glacier area (March 1980) obtained is 0.059 ± 0.002 km², equivalent to the 28.4% of the pre-event glacier area (Figure 5H), whereas the ice-debris deposits left within the 1980 pre-detachment glacier area (Figures 2, 5) cover an area of 0.024 ± 0.002 km².

Figure 6A shows the spatial distribution for the glacier elevation changes along the 2015 GNSS survey track when compared with the 2020 LiDAR survey. A mean elevation change rate of −0.74 ± 0.24 m a⁻¹ was observed, equivalent to a change of −3.65 ± 1.19 m for the 2015–2020 period. Maximum thinning values beyond −8.0 ± 1.19 m are observed in the central section of the glacier, coinciding with the widest portion of the glacier both in 2015 and 2020. As shown in Figure 6, the ice thickness changes show mostly negative values, indicating that ice thinning has prevailed over the glacier in the period 2015–2020. We can also observe that the upper area of the glacier shows more thinning than the lower area, possibly due to the existence of cleaner ice, opposite to the lower area which has a thicker debris cover, probably due to enhanced ablation at lower elevations. For the post-event conditions, an elevation difference of −30 to −50 m was observed at the margins of the lower half of Aparejo Glacier for both 1980–2015 and 1980–2020 periods (Figure 6B).

Figure 7A shows the spatial distribution of glacier thickness, based on a 2,588 m-long GPR profile collected on Aparejo Glacier. After applying the corresponding filtering of a minimum thickness value of 10 m (see Section 4.3), reliable glacier thickness data could be obtained for a profile length of 1,413 m, corresponding to 54.6% of the total length. The maximum thickness observed was 40 m on a clear reflector between the distances of 600 and 1,100 m, considering an ice velocity equal to 0.168 m/ns (Figure 7B), whereas the average thickness observed was 19.5 m. The highest thickness values, over 30 m, are concentrated in the medium third of the glacier. This value decreases slightly toward the glacier front and decreases on a more abrupt pattern toward the headwall.

Figure 8 illustrates a reconstruction of Aparejo Glacier right before and after its sudden detachment. Also, the 2015 and 2020 surface profiles are presented along with the glacier bed topography after the GPR data. The profile location is shown on Figure 7A. Before its detachment, Aparejo Glacier had an extension of 0.21 km² (Figure 5; Table 4). The margins near the glacier terminus reached an elevation of 3,420 m a.s.l., whereas the glacier’s length was about 1 km. The headwall position is observed at the same position as both before and after scenarios (1956 to 1967 and 1996 to 2020, Figure 5). The average surface slope of the glacier is estimated as 21°. According to GPR data (Figure 7), compared to the 1980 margin’s position and assumed elevation, the maximum pre-event thickness of the glacier is estimated in 77 m (Figure 8), whereas the average thickness is estimated as 62 m. This value allows us to estimate a pre-detachment glacier volume of 12.9 ± 0.6 × 10⁶ m³. We also estimate a 68 m elevation drop for the main fracture that started the sudden detachment. For the 0.06 km² glacier area left after the detachment (Table 4), we estimate an average thickness of 20 m based on the 1980 post-event photogrammetric model (Figures 5H, 6B), which allows us to estimate a post-detachment glacier volume of 1.2 × 10⁶ m³. Thus, the glacier detached volume is estimated at 11.7 ± 0.6 × 10⁶ m³, equivalent to the 90% of the glacier’s pre-detachment volume.

Supplementary Table S3 shows the results of the isotopic analysis of 26 samples shown in Supplementary Figure S1. The most depleted values of δD and δ¹⁸O are observed in samples from Glacier 51 (blue dot on Supplementary Figure S1A), with less depleted values on water samples collected downstream. The most enriched values are observed in the sample collected in a small lake (sample A15 on Supplementary Figure S1A) near the main stream, which may be affected by strong evaporation as discussed further in this section (Clark and Fritz, 1997; Spangenberg et al., 2007). The lake has been present since at least 1956, as revealed by Hycon aerial photos in Supplementary Figure S1B. The latter condition allows us to exclude the lake sample for further analysis.

Figure 9A shows the distribution of water stable isotopes values with respect to altitude, showing that no evident relationship of water isotope composition and altitude is present. Stream samples have homogeneous stable isotope values of water despite the sampling location, meaning that no important water-mixing is occurring downstream. The exceptions are the two highest samples, which values are similar to those of the Aparejo Glacier. Glacier 51 has a clearly distinguishable isotopic signature, much more depleted than any other water source from this sector. Considering that below 3,300 m a.s.l. there is no presence of ice, no clear relationship can be established. Nonetheless, snow samples collected likely show an isotopic depletion as altitude increases (Rozanski and Araguás, 1995).

Figure 9B shows the isotopic composition of all collected samples, which plot close to the Global Meteoric Water Line (GMWL), as defined by Rozanski et al. (1993), except for the one from the lake, not included in the plot. Three clusters of samples are recognizable, from more depleted to more enriched water stable isotope values: those from Glacier 51, the ones from the stream and those obtained at the Aparejo Glacier. Snow samples plot in a sparser way, showing higher variability, probably due to substantial differences in the deposition time of the different snowfall events, that could reflect different sources, environmental conditions and also post-depositional processes like sublimation.

Figure 4A shows the Daily Zero Isotherm fluctuations measured for February and March 1980 obtained from the coastal radiosonde station Quintero. For February 1980 we observed both cold and warm days with a minimum and maximum Zero Isotherm Altitude (ZIA) of 2,750 and 4,750 m a.s.l., respectively, whereas the ZIA average for 1975–1998 is estimated around 4,000 m a.s.l. As for the meteorological situation for the day of Aparejo Glacier detachment, 1 March 1980, we observe the presence of an upper-level trough crossing the Andes, as detected by the 500 hPa analysis and the descent of the ZIA (Figure 4A).

In February 1980 two important precipitation events took place, one in 9 February and the other between 19 and 23 February (Figure 4A). Both events are accounted for by the daily precipitation measured in El Yeso and Lagunitas stations. On the nearest weather station, El Yeso (2,486 m a.s.l., Table 1), the precipitation amount varied between 23 and 14 mm, respectively, whereas in Lagunitas station (2,760 m a.s.l., Table 1), the total precipitation was 14 mm for 9 February and 43 mm for the event of 19–23 February.

Figure 4B shows the accumulated February precipitation (liquid) for all available stations deployed in the western side of the Andes near the Aparejo Glacier. The areal average of liquid precipitation reveals February 1980 as the highest value (32 mm) recorded between 1978 and 2004. According to the Lagunitas station data, located 60 km to the NNW of Aparejo Glacier, which constitutes the oldest and most reliable mountain station in the Central Chilean Andes, the February 1980 precipitation is also the highest recorded within the period, followed by the 1963 February record.

As Figure 5 and Table 4 show, Aparejo Glacier has had an evolution from a relatively steady area of 0.13 km² for the 1956–1967 period, to a remarkably large growth of 0,08 km² until its pre-detachment extension of 0.21 km² was reached in 1980. Our results show a 200 m length reduction of the glacier from 1956 to 1967, and a 50 m wider geometry for the 1967 outline when compared to the 1956 outline. Interestingly, Figure 5B shows what resembles a bulgy-like geometry in the middle of the glacier for 1967. However, it was not possible to analyze the bulge elevation due to the lack of a DEM for 1967. For the 1967–1980 period, we observe a 190 m glacier front advance. What is more remarkable is the widening of the glacier at the 1967 terminus position. The latter would have varied from a few meters in 1967 to almost 200 m in 1980 (Figure 5C). Bulging before glacier detachments events have been reported for Kolka, Aru, Sedongpu and Flat Creek glaciers (Kääb et al., 2021). In the case of Aparejo Glacier this feature would have been accounted for 13 years before the sudden detachment event.

As for the pre-detachment glacier mass balance, although we do not have any data, Farías-Barahona et al. (2019) give an accurate assessment for the Echaurren Norte Glacier, located 11 km southwest of Aparejo Glacier. As part of their results, the authors compare the annual mass balance of the Echaurren Norte Glacier and annual precipitation at El Yeso station between 1975–2015 (Farías-Barahona et al., 2019, Figure 6). For the period 1975–1980 one El Niño event occurred in 1978 and a positive mass balance for Echaurren Norte Glacier was verified for the 5-year period, with annual precipitation slightly below 500 mm at El Yeso Station for all 5 years, with the exception of 1978, when the precipitation was 787.5 mm. Garreaud et al. (2019) (Figure 2) calculate an annual series of the Central Chile regional precipitation index (RPI). A total of 9 years with RPI over 100% compared to the average are accounted for within the 1956–1980 period, whereas 9 years qualify as droughts, and another 7 years with precipitation in the range of 80–100%. The latter agrees with the 9 weak-moderate-strong El Niño years identified for the same period according to the Oceanic Niño Index (NOAA, 2024). At the same time, temperature anomalies had a negative (cooling) tendency between 1958 and 1976 (Figure 3A in Barría et al., 2019). The tendency reverted just after 1976, leading to the only positive mass balance of Echaurren Norte Glacier in 1978. This idea is reinforced by the fact that in 1976–1977 the Pacific Decadal Oscillation (PDO) shifted from a negative to a positive phase (Quintana and Aceituno, 2012; Barría et al., 2019).

As for the bulging and glacier growth between 1967 and 1980, the geometry of the glacier lateral moraine (Supplementary Figure S3) may have played a role favoring its development. This feature presents a 30 m narrowing adjacent to the Aparejo Glacier frontal position on every existing outline, with the exception of the 1967 extent (Figure 5). In particular, the 1980 glacier outline has its terminus next to the narrowest portion in between the lateral moraines (Supplementary Figure S3). This configuration could have led the glacier to grow on a confined condition prior to its sudden detachment. It is worth noting the similarity of the latter with the pre-detachment scenario for Tsambagarav Glacier detachment (48.66, 90.75), Altai Mountains, Mongolia, after the 1987 outline observed by Kääb et al. (2021). In both cases the sinuosity of the lateral moraines would have had a similar role in the glacier’s evolution before their sudden detachment.

Although Masiokas et al. (2016) reconstruct a negative mass balance trend for Echaurren Norte Glacier for the 1940–1970 period, which would have stabilized soon after the beginning of the 1970 decade, considering the southern exposure of Aparejo Glacier, it is reasonable to assume that the local meteorological conditions prior to its detachment would have led to the glacier to duplicate its area on a 24-year span.

As for the possibility of past detachments of Aparejo Glacier, the strip lines evident on the 1956 aerial photographs in the valley downstream of the glacier (Supplementary Figure S1B) are extremely similar to the strip lines along the glacier deposit path after the 1980 detachment, which points to the occurrence of similar past events, perhaps even larger. In fact, during the Little Ice Age (Fernández-Navarro et al., 2023), the glacier must have been much larger, potentially having had repeated instabilities such as sudden large detachments. Nonetheless, careful field investigations (Jacquemart et al., 2022) are necessary to determine whether Aparejo Valley older deposits could correspond to previous detachments.

Similar to Kolka Glacier catastrophe (Kotlyakov et al., 2004; Huggel et al., 2005; Evans et al., 2009; Kääb et al., 2021), one atmospheric factor that could contribute to the collapse of the glacier is a warmer than usual environment. Analyzing the upper air temperature at 850 and 700 hPa obtained from the coastal radiosonde station Quintero (Figure 4A), this factor however, is discarded so that the monthly average for February 1980 was below the corresponding long-term average. Indeed, the days before the Aparejo Glacier detachment colder than average air temperatures were recorded, as reflected by ZIA values below average (see Figure 4A).

Considering the latter, precipitation arises thus as a main atmospheric factor that needs to be evaluated as a potential detachment trigger. Both summer precipitation events that took place in February 1980 (Figure 4A) occurred with a ZIA lower than average and below or at least within the altitudinal range of the glacier. This means that the precipitation on the glacier was most likely of solid type (snow). Nonetheless, after the precipitation events of February 1980, the ZIA rapidly rose above the altitudinal range of the glacier, thus favoring snow melt. No precipitation was recorded just before the Aparejo Glacier collapse at the few stations deployed in the central Chile Andes, except for the Lagunitas station located around 60 km to the north, that recorded 1.5 mm on 1^st March, with a ZIA of 3,658 m a.s.l. on February 29 that rose to 3,810 m a.s.l. on the 1st of March. According to the witnesses, snowfall on February 29th accumulated a 10 cm surface layer of wet snow on Aparejo Glacier and surroundings (R. Mujica, personal communication). The supraglacial ponds that the mountaineers observed while crossing the glacier, just before the event, were probably the result of extensive snow melt on the glacier.

These observations suggest that the triggering mechanism of the glacier detachment likely involved an extreme reduction of the basal drag due to high water saturation of the glacier bed. Aparejo Glacier appears to sit on a glacier bed composed primarily of weak subglacial till, and the slope on the lower two thirds of the glacier averages 7° (according to field observed data, after Marangunic, 1980). Snowmelt infiltration and warm precipitation due to a sudden increase in the zero-degree isotherm elevation would have provided the main source of infiltrated water, leading to enhanced water pressure at the glacier bed and, thus, reducing effective pressure (Jacquemart and Cicoira, 2021; Kääb et al., 2021). As for the main water pathways, according to witnesses and post-detachment footage (Supplementary Figure S4), Aparejo Glacier had a severe fractured texture on its surface. Thus, it is reasonable to assume that much of the melt water present just before the detachment, as evidenced by the surface ponds reported by the witnesses, reached the glacier bed facilitated by numerous crevasses, thus increasing the subglacial water pressure. Abnormal crevassing before failure is known to be a shared condition for other detached glaciers such as Kolka, the twin Aru glaciers, Tsambagarav, Sedongpu and Rasht (Kääb et al., 2021).

As for the abnormal crevassed condition for Aparejo Glacier before the detachment, although there is no known pre-event footage, besides what is shown in Figure 5 for the 1956 and 1967 conditions, it is worth noting the surge-like behavior of nearby Glacier 51. Indeed, during a field inspection on 12 March 1980, Marangunic (1980) found that the nearby glacier to the east, Glacier 51 (Figure 1), showed significant signs of surge-like instability, such as a heavily crevassed front and massive patches of freshly exposed ice along its entire length (Supplementary Figure S5). The prominent terminal moraine at Glacier 51, with over 100 m of elevation difference compared to its proglacial plain, probably helped containing its detachment. As Supplementary Figure S5 shows, it is remarkable how the surge-like behavior of Glacier 51 resulted in an overflow of glacier ice on both margins and on its front. What is most conspicuous, is the synchronicity of both phenomena, Aparejo Glacier detachment and Glacier 51 surge-like abnormality, within a distance of 1.5 km in the same basin. As Kääb et al. (2021) suggest, sudden glacier detachments could be seen as rare and extreme endmembers of surge-like glacier instabilities, thus, for both Aparejo Glacier and Glacier 51 to have exhibited a similar instability, a common local factor (beside meteorological conditions) must be present. Remarkably, the reconstructed pre-detachment surface topography of Aparejo Glacier (Figure 8) shows in the lower two thirds a thicker (bulgy) geometry than in the upper section of the glacier. This condition is typical of surge-like behavior (Jiskoot, 2011).

Regarding the geological role, we propose that the local geological conditions played a key role on the Aparejo Glacier detachment and, at the same time, on the surge-like glacier behavior of Glacier 51. As was described in section 2, Aparejo Glacier is in middle of the Aconcagua Fold and Thrust Belt and, on a more detailed scale, Aparejo Valley east margin is flanked by the El Diablo Fault. This configuration favors the near vertical disposition of the sedimentary strata on the headwall of Aparejo Glacier (Supplementary Figures S2B,C, S4) and Glacier 51, which is also coronated by Aparejo Peak (4,794 m a.s.l.), a massive gypsum body surrounded by sedimentary formations. On both glaciers at the headwalls of Aparejo Valley, the vertical drop from the highest summits to their respective accumulation zones is 500 to 900 m, whereas the horizontal distance to the accumulation zone lies in the 600 to 1,000 m range, resulting on a steep 40° slope over Aparejo Glacier and Glacier 51. At the same time, as Rodríguez et al. (2020) observed, the main recognized geological structures near Aparejo Glacier (El Diablo Thrust) are dominantly oriented parallel to the detachment path, favoring the preferential direction of the mass movement.

Considering the southern exposure of Aparejo Valley, plus the weakness of the sedimentary formations, mainly sandstones and fine-grained conglomerate, all above 3,500 m a.s.l, this favors the development of gullies above Aparejo and 51 Glacier, where snow avalanches are triggered, depositing large amounts of snow on the glaciers. Indeed, as Supplementary Figure S3 shows, snow avalanches originating above Aparejo Glacier can reach its terminus after traveling over 1 km. In this configuration, glacier accumulation is sustained mostly by snow avalanches, especially in cold and humid years, whereas in summer debris flows are predominant, allowing Aparejo Glacier to receive uninterrupted inputs of mass. This scenario poses a parallel process to what Jacquemart et al. (2020) found for Flat Creek Glacier, Alaska. In their study, the authors observed that the area upstream of the bulge identified on the glacier, before its sudden detachment, received water from a larger catchment than any other point of the glacier. In the case of Aparejo Glacier, instead of a larger catchment, we state that gullies above the glacier became main corridors for snow avalanches, debris-flows and, thus, an intense mass input both before and after its sudden detachment. A similar phenomenon occurs north of Aparejo Glacier in the Las Vacas Creek (−33.488, −70.044), Río Colorado basin, were successive debris-flows are triggered almost every summer, such as the episode of intense rainfall corresponding to the Zonal Atmospheric River (ZAR) of January 2021 (Sernageomin, 2021). The Las Vacas basin is a small watershed of 34 km² of which 29% of its upper area share the same topographic boundary as the Aparejo Valley. The fact that at both sides of the topographic boundary between Aparejo Peak and its nearest western height at about 4,200 m a.s.l., there is frequent evidence of debris flows and snow avalanches, this points to a particularly high accumulation regime for Aparejo Glacier and, likely as well for Glacier 51. This geological and topographic setting provides essential clues of the factors that contribute to trigger glacier instability.

As for the 1980 reconstructed geometry of Aparejo Glacier (Sections 5.1 and 5.4), the average width of 220 m coupled with the average slope of 21° (Figure 8) points to a loss of basal friction of about 90% based on the stability diagram (Figure 20) proposed by Kääb et al. (2021).

The position of the main three groups of samples in Figures 9A,B suggest a possible water mix from different sources, where the isotopic signature of the stream water could be the result of the mix between Aparejo Glacier and Glacier 51 meltwater, possibly along with other unidentified sources such as groundwater, according to Crespo et al. (2017). In their study, the most depleted waters originated from debris-covered glaciers, while the most enriched waters originated from rock glaciers. In this case, although both glaciers are debris-covered, Glacier 51 ice is older than the Aparejo Glacier detachment event, so that its isotopic signature could reflect the signal of older ice.

We propose that most of the remnant ice after the 1980 event (1.2 × 10⁶ m³) has been preserved and has become part of the present Aparejo Glacier. This is reasonable, considering that the Aparejo Glacier has a southern exposure with low solar radiation. Consequently, the remnant ice preserved in the upper section of the glacier after the detachment (0.059 km², Table 4) must have flowed downglacier, probably also sliding and avalanching down steeper slopes to the glacier’s mid and lower portions. It is also probable that snow accumulation due to precipitation and avalanches from steep slopes at both sides of the glacier contributed to fill the glacier basin. This transformation of snow into firn, and then into ice, would have easily occurred in a couple of decades, considering that the ice is probably near-temperate based on the relatively low elevation of the glacier.

This idea is reinforced by looking at the isotope secondary parameter of deuterium excess (d_excess) (Dansgaard, 1964). The d_excess of meteoric water mainly depends on the conditions during primary evaporation from the ocean (e.g.: relative humidity and sea surface temperature). This signature will be preserved as water falls as precipitation (solid or liquid) and is defined by the relationship: d excess = δ D − 8 δ 18 O

Figure 9C shows the d_excess values for all samples analyzed in this study. Glacier 51 samples show much higher values than all other samples, which in turn will reflect different environmental conditions at the moisture source during the evaporation process, as the moisture source should remain the same (i.e.: Pacific Ocean). Higher d_excess are linked to lower relative humidity (i.e.: colder air temperature conditions) (Clark and Fritz, 1997). On the other hand, Aparejo Glacier samples have lower d_excess values, likely reflecting higher relative humidity conditions over the ocean (primary moisture source), while stream samples probably reflect a mix of both older Glacier 51 and Aparejo ice melt.

The fact that the stable isotopic signature of Aparejo Glacier’s ice closely resembles that of the sampled snow above the glacier (Figures 9A–C), suggests that its temporal scale and temperature conditions at its deposition time are similar. Contrastingly, a clear difference of around −4‰ (−35‰) for δ¹⁸O (δD ‰) is found for Glacier 51’s ice close to the glacier front, suggesting the presence of older ice. Therefore, Aparejo ice is probably younger than the one of Glacier 51, supporting the recent glacier regeneration which has occurred at Aparejo Glacier.

During the 1980s and 1990s particularly humid years occurred in central Chile, related to a higher frequency of El Niño events, linked in turn to a positive phase of the Pacific Decade Oscillation (PDO) (Quintana and Aceituno, 2012). As Farías-Barahona et al. (2019) observed, after analyzing the mass balance data from the relatively close Echaurren Norte Glacier,² located 11 km southwest of Aparejo Glacier, and correlating it with the Multivariate ENSO Index (MEI) data for a time scale spanning from 1975 to 2015 (Figure 5 in Farías-Barahona et al., 2019), we can see the existence of a relationship between both time series. As the authors reveal, there is a strong relationship between the mass balance of Echaurren Norte Glacier and the ENSO index until the early years of the 2000–2010 decade. In fact, through the superposition of ENSO data, it is possible to identify five strong El Niño episodes: 1982, 1987, 1991, 1997, and 2002. All of them correspond to winter mass balance values higher than 2 m w.e. at Echaurren Norte Glacier (Farías-Barahona et al., 2019).

The first evidence of Aparejo Glacier’s extent after the 1980 detachment corresponds to the aerial photographs of 1996, when it had an area of 0,128 km², thus having gained an area of 0.069 km² with respect to the remnant area of 0.059 km² immediately after the detachment. In terms of volume, we can assume that it also increased after the 1980 detachment, at least until 2002, during which five winters with high precipitation were recorded in central Chile, linked to El Niño episodes, as mentioned above. After 1996, area changes have not been substantial (Figure 5; Table 4), and we may assume that glacier volume has shown a moderate increase until 2010, when a megadrought started to affect a major area of Chile. Therefore, the main ice regeneration process at Aparejo Glacier probably took place during the 22 year-period 1980–2002, helped by the strong El Niño episodes. Notably, this time span is remarkably similar to the documented growth of Crater Glacier, Saint Helen Volcano, USA (Walder et al., 2004). Nonetheless, the latter would’ve grown highly influenced by a lava dome emplacement within the post 1980 collapse crater of Saint Helen Volcano.

Starting in 2010, a megadrought began to affect central Chile (Garreaud et al., 2017), partial evidence of which is the average glacier thinning of 3.65 m detected during the 2015–2020 period. In 2015, a maximum ice depth of 40 m was derived from the GPR data in the upper area of the glacier (Figure 7), in contrast to the maximum ice depth of 73 m inferred for the original glacier before the slide (Marangunic, 1980). Thus, in the present scenario of drier conditions, in addition to climate warming in the Central Andes of Chile (Burger et al., 2018), most likely Aparejo Glacier will continue to lose mass in the form of ice thickness depletion, regardless if its glacier front stays at the same position.

This current negative mass balance scenario allows us to estimate as very unlikely the risk associated to a new detachment of Aparejo Glacier. In fact, in terms of its mechanical stability, a continuous decrease of ice mass will diminish the driving forces of a sudden detachment scenario, so the glacier is and will be out of the “instability envelope” (Jacquemart and Cicoira, 2021; Kääb et al., 2021). Most probably then it will remain stable in the upcoming decades, unless climate trends reverse, with either wetter winters and/or cooler summers, both of which are not probable in the near future (IPCC, 2021). An additional factor that produces glacier mass loss at Aparejo Glacier, and thus reduces the risk of glacier instability, are the more frequent and highly fluidized debris flows produced especially during intense summer precipitation events, as is observed for the 2020 topography in Figure 8, which can erode and melt a relevant mass of ice. On that basis, Google Earth imagery also shows continuous debris deposits originating from the upper gullies located between the headwall strata, similar to what is shown in Figure 5G.