the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 22 Oct 2025
The Anthropogenic Influence on Glacier Retreat in Central Chile
Abstract. Glaciers in the Andes mountain range have retreated since the mid-20th century. This change has been attributed to climate change and the effects of local pollution. Some glaciers subjected to similar meteorological conditions and the same influence of climate change are found to exhibit significantly different retreat rates, which cannot be explained by climatological factors alone. In the Maipo River basin, located in central continental Chile, two glaciers with similar climatic and geomorphological characteristics, the Paloma Oeste Glacier (POG) and the Bello Glacier (BG), exhibit significantly different levels of ablation. We implement two multivariable regression models, one per glacier, to identify the importance of specific climatic and anthropogenic variables in glacier surface area retreat. These models incorporate a temperature-related variable and precipitation (indicative of climate change), surface-level Black Carbon concentration (BC, indicative of anthropogenic activity) and the large-scale climate indices PDO and Niño 3.4 (related to climate variability) as covariates. The results indicate that the glacier surface change is more sensitive to the surface BC concentration and the Niño 3.4 index than to precipitation, PDO, or temperature for both glaciers. However, since the surface BC concentration in POG is more than 40 times higher than in BG, the area retreat is significantly higher in POG than in BG. Between 2000 and 2020, 49 % of the area retreat of POG is explained by BC pollution, while 97 % of BG’s retreat is explained by climatic effects (climate change and climate variability). Furthermore, when analyzing the causes of glacier retreat in POG before and during sustained drought conditions, often referred to as the Central Chile Megadrought (2010–2020), we found a change in the relative importance of BC surface concentrations. Before the Megadrought, BC is identified as the leading cause of glacier retreat in POG, accounting for a 53 %. However, the climatic effects (61 %) on glacier retreat during the Megadrought become more relevant than the impact of BC (39 %). These results highlight the spatiotemporal varying influence of climatic and anthropogenic factors on glacier retreat, emphasizing the significant contribution of climate change, particularly during sustained drought conditions.
- Preprint
(9264 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 18 Jan 2026)
- CC1: 'Comment on egusphere-2025-3715', Yulan Zhang, 28 Oct 2025 reply
-
RC1: 'Comment on egusphere-2025-3715', Anonymous Referee #1, 10 Dec 2025
reply
This paper presents an extensive analysis of causes of glacier area retreat over two morphologically and climatically similar glaciers in Central Chile with different retreating rates. The differences in retreating rates are attributed to reduced ice albedo due to higher black carbon concentrations in the fast-retreating glacier, which is located nearby anthropogenic mining activities. The authors build multilinear regression models to estimate annual glacier area change based on climatic and anthropogenic variables. Most of the analyses presented are robust, and the presentation of results is clear (with some exceptions which I elaborate below). The paper is generally well written, well structured, and easy to follow. However, I have important concerns regarding the novelty and framing of the study, as well as the presentation and interpretation of some of the results.
In its current form, the paper does not sufficiently articulate the novelty with respect to the study from Cereceda-Balic et al. (2022) titled “Understanding the role of anthropogenic emissions in glaciers retreat in the central Andes of Chile”, which is a very similar title to the title of this paper. Both studies compare the glacier retreat rates from the same glacier (Bello Glacier), to the retreat rates from a glacier located near a mining area (Olivares Alpha Glacier in Cereceda-Balic, and Paloma Oeste Glacier in this study, which are located a few km away from each other). The results in terms of attribution of glacier retreat to black carbon compared to climatological factors are therefore very similar between the two studies.
In my opinion, this paper presents two main novelties that the authors need to put more value on, with some caveats. The first one is the incredibly high performance of a simple multilinear regression model in predicting glacier area variability based on some rather simple climatic variables and indices (explaining up to 95% of glacier area variability). The second one is the shift from black carbon dominated glacier area change to climate dominated glacier area change with the onset of the Chilean Megadrought. The current title and main messages of the paper do not sufficiently articulate these novelties. Furthermore, in its current form the paper focuses on glacier area retreat alone and lacks discussion on the implications of the findings beyond the impact of anthropogenic activities on glacier area retreat (for instance, hydrological implications). A revised paper would be more suitable within the aims and scope of another EGU journal such as The Cryosphere and I therefore do not recommend publication in HESS in its current form.
Main comments:
Data availability: The authors should comply with the journal data policy (https://www.hydrology-and-earth-system-sciences.net/policies/data_policy.html), making data available or clearly explaining why that is not possible. In any case the data for reproducing the figures must be made available. There is no indication of where the meteorological data can be found, other than “Data provided on request” in the data availability section. Authors obtained records for PDO and El Niño from NOAA but only point to the general NOAA website, and not the actual data records. The same occurs for atmospheric composition data, pointing only to the Copernicus Atmosphere website but not the specific dataset used.
Section 2.2.2: The use and postprocessing of meteorological data is unclear. The authors present a table with the weather stations in the area (Table 3), but then it seems that only two weather stations are used (Lines 145-146), while additional weather stations were used for gap filling. There are no details about this gap filling. What data series were filled, how, and how many gaps were there? In line 143-144 “precipitation data were extrapolated from nearby meteorological stations”; what does this mean? How was it extrapolated? Same with temperature in line 148 “extrapolating to the glacier elevation” does this mean lapse rates were applied, and what values were applied?
Some of the methods outlined in Figure 2 are not clearly described in the text (e.g., temporal adjustment, homogenization, visual validation, outlier detection-correction, manual checking).
L272: “At the current rate of retreat, POG could disappear in just over 10 years.” This statement should be removed, or evidence should be shown. Where does the estimation of 10 years come from? Judging from the trend lines in Figure 3, it does not seem like either glacier will disappear within 10 years, and this is assuming the Megadrought will continue.
Throughout the results and discussion section, the authors write statements with trend values based on the 20-year time series available. In many cases, the trend values presented refer only to the 10 years of Megadrought period from 2010 to 2020. The values of these trends are therefore very high, and large affected by the Megadrought period. The authors should apply caution throughout the manuscript in presenting these as significant trend values, given that 10 years is not enough for climatological trends to be significant, especially as they could easily be reversed if the Megadrought would stop. Instead, these should be presented as rates or changes within the 10-year period, with clear explanations that these are clearly affected by the Megadrought period and not necessarily long-term climate trends. Examples of this are in line 281-289.
Figure 4: These should be presented as time series (line plots with time on the x-axis), instead of boxplots.
I believe there is an error in the calculation of the Mean Absolute Percent Error (MAPE). Based on Figures and Tables, I think the authors did not multiply the MAPE by 100, to make it a percentage. This is especially clear in Figure 9, where a RMSE of 0.75 km2 corresponds to a 4% error. With the area of the glacier being < 4 km2, I think this is wrong. This then affects the results and discussion presenting the model with tiny errors, such as in line 349 (“The MAPE remains below 0.1%”).
It is an incredibly good result that the multilinear regression models explain up to almost 97% (Fig. 7a) of the variability in glacier area. However, it seems difficult to believe that a simple model can explain such a high variability of the glacier area. The temperature and precipitation data are not observed on the glacier but extrapolated from nearby stations. The two climate variables used are large-scale climate indices which I assume have a coarse resolution (not stated in the paper). The Black Carbon variable has a 0.1 degree resolution. All seem too coarse to capture the effect of these variables at the location of the glaciers, given their small sizes. Nevertheless, the authors should clearly discuss this almost perfect predictive performance. Is this performance possible to extrapolate to other glaciers in the region? If so, that would be great and could lead to a follow-up study, or a reframing of this study, to investigate regional variability in glacier area change with a simple model. If not, it could be that the model is overparameterized for these glaciers. Or perhaps the fact that these are relatively small glaciers has an influence on the results? Further detailed discussions on this matter should be provided. Furthermore, the dependent variable (observed glacier area) must have some uncertainty associated to it, but this is also not discussed or presented.
Other comments:
- L53: “changes experienced by the Cryosphere”. Does this mean globally? Then a few more references of global studies should be included.
- L73: “does not appear to be explained by glaciological factors alone”. What are the glaciological factors? Do the authors mean climatological?
- L80: -1.2% AND -0.6% of what? Melt rates or glacier area?
- L103: Isn’t there a more modern inventory than the one in 1979? With strong glacier retreat, I would guess these numbers may have changed.
- Table 1: Would be good to add area of the glacier as a column.
- L149-151: Please use Positive Degree Days (PDD), instead of DTbCero. PDD is a more common variable in climate studies, and is even the variable used in the two papers that are cited to justify the use of DTbCero (e.g. Vincent 2002 and Wiltshire 2014).
- Table 2: I think this table could move to an appendix or Supplement.
- L189: The use of the word “evaluate” here is not clear what is referring to.
- L236: Is leaving one year out enough for a cross validation in this case? As the data series are 20 years.
- Figure 3: Remove [yyyy] from the labels. Regarding the colourbar of years: while the current colour scale is useful to identify single years, I think here it would be more useful or impactful to have a continuous colour scale that shows how the glacier is retreating through time.
- Table 5: Temperature should be PDD and not temperature, and the value should therefore be negative. Otherwise the positive correlation indicates that more temperature leads to larger glacier area. Please also make cm2 and Km2 a superscript for 2.
- Figure 5: Are the R2 values or RMSE for these models only presented in Figure 9? If so, I suggest combining these figures.
- Table 6: Should 2004 be 2007 in the middle bottom of the table?
- L355: The difference between standardized and unstandardized is not entirely clear to the reader, making the difference between Figure 6 and 7 not clear either. What is the difference between predicted and simulated glacier area?
- L373: What is an “equivalent decrease in winter temperatures” compared to an increase in BC? As these variables are difficult to compare, equivalent here is ambiguous.
- Figure 9: Does KGE make sense for this evaluation?
- Figure 10: I think this is an interesting impactful figure and should be a more focal point of the study.
Citation: https://doi.org/10.5194/egusphere-2025-3715-RC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 210 | 157 | 19 | 386 | 11 | 14 |
- HTML: 210
- PDF: 157
- XML: 19
- Total: 386
- BibTeX: 11
- EndNote: 14
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
MS title:The Anthropogenic Influence on Glacier Retreat in Central Chile
MS NO.: egusphere-2025-3715
This study focuses on the analysis of melting of the Paloma Oeste Glacier (POG) and the Bello Glacier (BG) in the Maipo River basin, central Chile. It finds that the surface changes of both glaciers are more sensitive to black carbon concentration and the Niño 3.4 index than to precipitation, PDO, or temperature. Additionally, POG shows a more significant retreat due to its higher black carbon concentration, while the impact of climatic factors on glacier retreat exceeds that of black carbon during drought periods. Overall, the structure is coherent and the analysis is reasonable. However, one big question for this paper is: How does the author define "anthropogenic influence"? Is black carbon (BC) intended to serve as an indicator of anthropogenic activities in this region? It is argued that the paper’s arguments can only be meaningfully discussed if the author provides a clear definition of this term. Without such a definition, BC cannot reasonably be regarded as the sole indicator of anthropogenic activities affecting glacier melting.