Groundwater provides temporary resilience to alpine catchments in glacial retreat

doi:10.21203/rs.3.rs-2204053/v1

Download PDF

Article

Groundwater provides temporary resilience to alpine catchments in glacial retreat

https://doi.org/10.21203/rs.3.rs-2204053/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Surface runoff from melting alpine glaciers increases until the glaciers reach a condition called “peak water” beyond which surface runoff decreases as ice volume diminishes. Recent research indicates that a fraction of this meltwater recharges mountain-block aquifers thereby providing vital support to baseflow in alpine streams and perennial flow from alpine springs. However, the response of baseflow to deglaciation has not been thoroughly quantified. Combining statistical timeseries analysis of discharge data with the delayed flow index (DFI), we show that the proportion of baseflow in post-peak water discharge has increased compared to pre-peak discharge in three glacial catchments in the Swiss Alps. The timing of peak water is not concurrent with the baseflow response. The increase in baseflow is sustained for 20 to 40 years after the occurrence of peak water; however, this is a transient response. In fact, one catchment has lost baseflow support and is now responsive to seasonal precipitation events. Given the slower response times of groundwater, baseflow will eventually decrease in these catchments making them more responsive to precipitation events and less resilient to fluctuations in precipitation once the ice is gone. These conditions will have severe impacts to alpine streams and springs and their ecosystems.

Earth and environmental sciences/Hydrology

Earth and environmental sciences/Climate sciences/Cryospheric science

Earth and environmental sciences/Climate sciences/Hydrology

Quantifying the impact of melting alpine glaciers on mountain groundwater systems is a new frontier in the study of alpine glaciers¹. Our fundamental understanding of the importance of alpine glacial meltwater on mountain-block recharge (MBR) and groundwater processes beyond the glacial catchment continues to evolve, but large uncertainties remain^2,3,4,5. For example, it remains unclear how glacial ice loss is impacting baseflow generation in streams and perennial flow from alpine springs. The hydrology and hydrogeology of glaciated alpine catchments have changed dramatically since the 1960s due to increased rates of glacial retreat^1,6. Thus, it's imperative that these knowledge gaps be addressed quickly since many alpine glaciers are rapidly retreating. Baseflow is defined as that portion of streamflow that is supported by delayed sources of water⁷ and it is especially critical in maintaining flow in streams during dry months or droughts. In alpine catchments, delayed sources of water may include water stored in soil, shallow aquifers in unconsolidated glacial sediment such as till and talus, and deep aquifers in bedrock^5,8,9,10,11. During the summer melt season, streamflow is an integration of different sources of water from different runoff mechanisms including surface runoff (overland flow) from glacial melt, shallow subsurface flow through soil, and groundwater. During the remainder of the year, baseflow becomes increasingly important at maintaining flow in the stream^2,3,4.

As alpine glaciers retreat, the volume of surface runoff from meltwater increases until the catchment reaches a state known as peak water¹². The peak-water conceptual model states that freshwater withdrawals can continue in a renewable fashion so long as withdrawals do not exceed recharge rates or replenishment rates¹³. As applied to glacial catchments, the volume of surface runoff must decrease after reaching the peak-water tipping point^12,14,15 since glacial ice is not being replenished. The volume of surface runoff from meltwater in a post-peak water state will be much lower than it was prior to the onset of glacier retreat or pre-peak water state¹². However, it’s unknown to what extent or how long baseflow will support streamflow from these catchments as the catchment transitions to an ice-free state.

The terms resistant and resilient are traditionally used in ecological studies^16,17, but they are also useful in assessing the responses of hydrological/hydrogeological systems to climate change¹⁸. Resistant implies that the system does not change due to stressors^16,17. In a hydrological context, resistant means that the system does not change due to stressors such as reductions in recharge or a loss of recharge associated with climate change¹⁸. Resilient, in comparison, implies that the system may change in response to a disturbance, but it will return to a state in which it has approximately the same function as it had in the pre-disturbance state¹⁷. Hydrological systems rarely show resistance to stressors, but they often show wide variability in resilience. The alpine glacio-hydrological system is no exception; it is not resistant to the effects of losing ice and it’s unknown to what extent these systems are resilient to the same stressors. Despite recent advances in quantifying recharge from alpine glacial meltwater^2,3,4,5,19, the baseflow response to deglaciation in alpine catchments remains largely unknown because there have been few studies focused on changes in baseflow as the glaciers retreat^3,4 and the peak-water conceptual model does not explicitly account for changes in baseflow¹². Consequently, we do not know what processes or which reservoirs of water provide resiliency in high-alpine catchments or how long these catchments may show resiliency.

Identifying the baseflow component in streamflow from alpine glacial catchments is complicated because there may be several reservoirs of delayed flow present in the catchment^5,8,9,10,11. Each reservoir has a characteristic hydraulic response time that is defined as the time it takes for discharge from the aquifer to respond to a disturbance^20,21, for example, a change in recharge. Groundwater responses always lag the disturbance. Lag-times can range from a few days to several decades or longer. Shallow aquifers typically have shorter response times than deep aquifers and/or bedrock aquifers although there are exceptions. The “true” hydraulic response time of the catchment is a multiplexed signal of the different response times²⁰. In this study, we investigate the groundwater response to alpine glacial melt by comparing changes in post-peak baseflow to pre-peak baseflow.

One hypothesis is that recharge from glacial meltwater is insignificant (except to shallow aquifers), therefore groundwater does not provide substantial resilience in a post-peak water state. Recent research shows that the amount of recharge attributed to glacial meltwater is highly variable and may be greatest near the runoff channel^2,3,4. However, other research shows that a substantial fraction of MBR is sourced from glacial meltwater and it supports flow within the mountain block^5,22. We test a competing hypothesis based on the latter which states that groundwater recharge from glacial meltwater is not insignificant, consequently baseflow provides short-term resilience to the alpine catchment in a post-peak water state. We test the hypothesis using the delayed-flow index (DFI) method²³ since it can, in principle, identify the different groundwater reservoirs in the catchment without the aid of additional field measurements used to identify the exact locations of groundwater stores²⁴. The DFI method is based on the local minimum method^25,26 where an algorithm checks a specific day of a hydrograph to see if it contains the lowest discharge in a pre-determined window size including adjacent days on either side of the specific day (see SM). The window then moves on to the next day in the hydrograph and repeats the calculation. DFI values are calculated for increasing window sizes, also called block length numbers (BLEN; reported in days)²³. When the BLEN is small (BLEN < approximately 10 days), the algorithm selects minimum discharge points over very short windows of time. These minimum values tend to be associated with larger and more variable discharge values with short delays representing groundwater from aquifers with perhaps limited storage, e.g., shallow aquifers in glacial sediment. When the BLEN is large (BLEN > approximately 40 days), the algorithm selects the lowest discharge point occurring over a much larger window in time, thereby excluding higher minimum discharge values and ignoring large fluctuations in minimum discharge²³. As a consequence, the delayed flow time series is much lower and less variable than the original streamflow time series and usually represents contributions of groundwater from aquifers with greater storage, e.g., bedrock aquifers with very long delays. Therefore, DFI values should decrease with increasing BLEN because larger BLENs are associated with reservoirs having increasingly attenuated event-scale responses.

Characteristic delay curves (CDCs) and breakpoints (BPs) identify different reservoirs (components) of delayed water that contribute to streamflow²³. The CDC is created by plotting the DFI (ranging from 0 to 1.0) against the BLEN²³. BPs are natural inflection points that are present in the CDC. The areas between consecutive BPs define the different delay classes of water. For example, the area between DFI0 (BLEN = 0) and DFI1 is called a short delay class of water, DFI1 and DFI2 is called the intermediate delay class, DFI2 and DFI3 is called the long delay class, and DFI3 + is called the baseline delay class²³. The baseline delay class is associated with the largest BLEN and is likely representative of groundwater from deeper aquifers. We calculate a DFI for each BP which represents the proportion of discharge associated with that delay class divided by the total discharge. Therefore, the DFI at each BP ranges from 0 to 1; the sum of DFI for all BPs equals total discharge. If DFI for a given BP increases post-peak as compared to pre-peak water, then this indicates that more baseflow is contributing to the stream post-peak water. If our competing hypothesis is true, then we expect to see an increase in the proportion of baseflow in a post-peak water state.

After rigorously screening 800 global glacial catchments (SM and Fig. S1), three catchments in the Swiss Alps were selected for this study: Glatt Firn, Rottengletscher (Rhone Glacier), and Roseggletscher (Fig. 1). The glaciers in these catchments have retreated considerably since the 1950s; Rottengletscher catchment (39.33 km²) lost 8.71 km² of ice, Roseggletscher catchment (66.19 km²) lost 13.5 km², and Glatt Firn catchment (20.96 km²) lost 3.84 km², representing 22%, 20%, and 18% of ice cover from 1850 to the early 2000s, respectively (Table S1). The geology of the catchments, change in glacial cover through time, and types of geological material that the streams cross in the catchments are provided in Figs. S3, S4, and S5. The peak-water year was calculated for each of the three catchments by plotting annual discharge through time and then applying an 11-year moving average of annual discharge to rule out long-term atmospheric oscillations⁸. The peak-water year was then used to divide the stream discharge, temperature, and precipitation time series into pre-peak water and post-peak water states (Fig. 2).

The peak-water years are 1990, 1996, and 1986 for Rottengletscher, Roseggletscher, and Glatt Firn, respectively. The Glatt Firn catchment reached peak-water nearly a decade before Roseggletscher catchment. Peaks in total annual stream discharge (m³ s^− 1) are not concurrent with peaks in precipitation for Glatt Firn and Rottengletscher catchments (Fig. 2, Tables S2, S3), thereby ruling out that hydrograph peaks are simply responses to precipitation events. However, peaks in total annual stream discharge for Roseggletscher are somewhat synchronous with peaks in precipitation that occurred in all three catchments in 1998 (post-peak water state). The Roseggletscher catchment contains a very large proglacial lake (0.39 km²; see Fig. S2) that likely attenuates the natural seasonality of discharge from the catchment. However, the data indicate that the Roseggletscher catchment is more responsive to precipitation events than the other catchments in a post-peak water state (Fig. 2, Table S4). The other two catchments are not as responsive to precipitation, instead increases in peak annual discharge are better explained by rising temperature and consequent release of glacial meltwater (Fig. 2).

The CDCs for Rottengletscher and Glatt Firn show dramatic differences pre-peak and post-peak water for longer BLENs despite being geographically close to one another (Fig. 3). However, there is a sharp positive transition in the “Post DFI – Pre DFI” variable (right vertical axis and bar chart in Fig. 3) after a BLEN of 29 days in Rottengletscher and 18 days in Glatt Firn. This indicates that a larger proportion of long-delay class flow is present in streamflow post-peak water as compared to pre-peak water. In other words, the proportion of baseflow has increased for both catchments post-peak water, yet the two catchments show different responses. In 1963, the pre-peak discharge from the Rottengletscher catchment was approximately 950 m³ s^− 1 and subsequently increased to approximately 1100 m³ s^− 1 in 1990 (peak-water year; Fig. 2, Table S5). Discharge has not yet reached 1000 m³ s^− 1 at the time of this writing, but it is decreasing gradually with time. Portions of Rottengletscher have been covered with white blankets since approximately 2006 or 2007 to slow the melting process and this has likely delayed the reduction in discharge from the catchment. In comparison, post-peak water discharge from Glatt Firn has fallen below that of pre-peak water discharge; the discharge was slightly higher than 550 m³ s^− 1 pre-peak water and it is now slightly lower (Fig. 2, Table S6). Recall that Glatt Firn reached peak-water sooner than the other catchments and its discharge has since fallen below pre-peak water levels.

Baseflow supports a greater proportion of discharge from Rottengletscher in a post-peak water state and that baseflow increase is primarily sourced from long-delay sources of water (higher BLENs). Specifically, the contribution from the BLEN of 29 has increased by 52% and the contribution from the BLEN of 43 has increased by 48%. Glatt Firn, in comparison, exhibits a characteristic peak-water curve where its post-peak water discharge has fallen below that of pre-peak water. However, the long-delay sources have increased substantially in Glatt Firn as well; the contribution from the BLEN of 32 has increased by 81% and the contribution from the BLEN of 44 has increased by 54%. Without these sources of baseflow, the discharge in both catchments would have decreased much more rapidly and by a larger magnitude. At the moment, baseflow is essential in maintaining water in the stream, but this is, unfortunately, a transient condition. Once the glacial ice is gone, that source of recharge is gone as well.

The CDC for Roseggletscher is quite different than the CDCs for the other two catchments. The “Post DFI – Pre DFI” for Roseggletscher is negative for BLENs greater than 40 days (Fig. 3, Table S7) indicating that contributions of groundwater associated with longer delay classed (BLEN > 41) have decreased rather than increased post-peak water. The Roseggletscher catchment is not located near the other two catchments, but it is developed in a similar geologic setting. However, the peak-water year for Roseggletscher occurred a decade later than the other two catchments and streamflow decreased more rapidly after peak-water (Fig. 2). Streamflow appears to become responsive to precipitation after 2007 (Fig. 2); yet there is a small, short-lived increase in the contribution from the BLEN of 41. One plausible inference is that the decrease in water from long-delay class simply indicates that the Roseggletscher catchment has limited “deep” storage compared to the other catchments since deeper aquifers tend to be associated with longer delay classes and larger BLENs. Consequently, this reservoir of water is insufficient to continue to support post-peak water streamflow. In the case of Roseggletscher, post-peak water discharge depends on short-delay class stores of water that are more responsive to atmospheric fluctuations.

Our data show that the proportions of baseflow in streamflow have increased post-peak water, but what are the response times of the groundwater systems to loss of ice? Providing an exact response time of these groundwater stores to a loss of recharge is difficult without the aid of a numerical model. In addition, the BLEN does not equal groundwater residence time or groundwater response time; these three variables are distinct from one another. However, it took 40 years for the post-peak water discharge from Roseggletscher to fall below pre-peak water discharge (Fig. 2). Streamflow from the Roseggletscher catchment today appears to be very responsive to precipitation events implying that it has limited resilience. This is not true for the other catchments. The peak contribution from the baseline delay class water occurred 12 years after the peak-water year in Rottengletscher and 16 years after the peak-water year in Glatt Firn (Fig. 4). In comparison, the peak contribution from the same delay class occurred 18 years after the peak-water year in Roseggletscher. The response from Rottengletscher must be interpreted carefully since portions of this glacier have been covered with blankets since 2006 to delay melting. Regardless, these findings show that peak baseflow is not concurrent with peak water. In addition, different stores of delayed water become increasingly important at sustaining surface flow as glacial ice melts in a post-peak water state. If our inferences are correct, it may take 2 to 4 decades for the alpine aquifers to respond to the initial change in recharge thereby providing some transient resilience. This finding supports recent research in Glacier National Park, Montana, USA where discharge from alpine springs was found to contain the stable isotopic fingerprint of glacial melt despite the complete loss of glacial ice in some catchments⁵.

We infer that the hydraulic response times of these alpine groundwater systems to the loss of recharge from glacial meltwater are likely on the order of 2 to 4 decades indicating that these catchments have short-term resilience in a post-peak water state, but they do not have substantial long-term resilience. In fact, once the groundwater stores become depleted, groundwater levels will decrease and the streams will become more responsive to precipitation events implying that the catchment will not function as it did in a pre-peak water state (i.e., the catchment has lost hydrologic resilience). Groundwater delays this response. Ultimately, the sources of groundwater that support these increases in baseflow are transient⁴; they will not be able to support discharges of the same magnitude as that observed in a pre-peak water state. With respect to our hypothesis, contributions of baseline and long-delay class groundwater are primary controls on the overall response to ice loss and are therefore vital to maintaining flow in the alpine stream in a post-peak water state. These catchments have limited short-term resilience to ice loss, but streamflow will gradually become more flashy (responsive to precipitation events). As groundwater stores become increasingly depleted, the streams may transition to an intermittent or ephemeral condition in the future with critical implications for the future integrity of aquatic ecosystems in these alpine glacial catchments as they transition to an ice-free future.

Acknowledgments: We thank Rene Paul Acosta for his assistance with locating, downloading, and analyzing the climatological data and with code development. We also thank J.B.M. dissertation committee members Nathaniel “Nat” Lifton (EAPS, Purdue University) and Darryl Granger (EAPS, Purdue University) for their valuable feedback on the dissertation chapter that lead to the development of this paper.

Funding: Funding for this research was provided by the National Science Foundation Grant (EAR 1904075 and 1904159). Additional support was provided by a Consortium of Universities for the Advancement of Hydrologic Science (CUAHSI) Pathfinder Fellowship, an Indiana Space Grant Consortium (INSGC) Graduate Student Grant, a Hydrologists Helping Others (H2O) grant and graduate student research grants through the Department of Earth, Atmospheric and Planetary Sciences at Purdue University.

Author contributions: M.D.F. and T.L.H. secured funding for the project. J.B.M., M.D.F., T.L.H., and A.K.L. conceived and planned the project. J.B.M. developed the codes to analyze the dataset and produce the figures. J.B.M. and M.D.F. wrote the paper. J.B.M., M.D.F., T.L.H., and A.K.L. reviewed and edited the paper. All authors contributed to improvement of this paper.

Competing interests: Authors declare no competing interests.

Data and materials availability: The data and all applicable codes will become available using Purdue PURR (https://purr.purdue.edu/). A doi is pending as data are embargoed per data management plan. In the meantime, Jordyn Miller will provide access to datasets and codes upon request.

M. Vuille, M. Carey, C. Huggel, W. Buytaert, A. Rabatel, D. Jacobsen, A. Soruco, M. Villacis, C. Yarleque, O. E. Timm, T. Condom, N. Salzmann, J.-E. Sicart, Rapid declines of snow and ice in the tropical Andes – Impacts, uncertainties and challenges ahead. Earth-Science Reviews. 176, 195–213, doi:10.1016/j.earscirev.2017.09.019 (2018).
A. K. Liljedahl, A. Gädeke, S. O’Neel, T. A. Gatesman, T. A. Douglas, Glacierized headwater streams as aquifer recharge corridors, subarctic Alaska, Geophys. Res. Lett., 44, doi:10.1002/2017GL073834 (2017).
L. D. Somers, J. M. McKenzie, S. C. Zipper, B. G. Mark, P. Lagos, M. Baraer, Does hillslope trenching enhance groundwater recharge and baseflow in the Peruvian Andes?, Hydrological Processes, 32, 318–331, doi:10.1002/hyp.111423 (2018).
L. D. Somers, J. M. McKenzie, B. G. Mark, P. Lagos, G. C. Ng, A. D. Wickert, C. Yarleque, M. Baraër, Y. Silva, Groundwater Buffers Decreasing Glacier Melt in an Andean Watershed—But Not Forever. Geophys. Res. Lett. 46, 13016–13026, doi:10.1029/2019GL084730 (2019).
J. B. Miller, M. D. Frisbee, T. L. Hamilton, S. K. Murugapiran, Recharge from glacial meltwater is critical for alpine springs and their microbiomes, Env. Res. Letters, 16, 064012, doi:10.1088/1748-9326/abf06b (2021).
R. Hock, Glacier melt: A review of processes and their modeling, Progress in Physical Geography, 29, 362–391, doi:10.1191/0309133305pp453ra (2005).
F. R. Hall, Base-flow recessions – A review, Water Resources Research, 4(5), 973–983 (1968).
C. Andermann, L. Longuevergne, S. Bonnet, A. Crave, P. Davy, R. Gloaguen, Impact of transient groundwater storage on the discharge of Himalayan rivers, Nature Geoscience, 5, 127–132, doi:10.1038/NGEO1356 (2012).
J. W. Roy, M. Hayashi, Multiple, distinct groundwater flow systems of a single moraine-talus feature in an alpine watershed, Journal of Hydrology, 373, 139–150, doi:10.1016/j.jhydrol.2009.04.018 (2009).
L. D. Somers, R. P. Gordon, J. M. McKenzie, L. K. Lautz, O. Wigmore, A. Glose, R. Glas, C. Aubry-Wake, B. Mark, M. Baraer, T. Condom, Quantifying groundwater–surface water interactions in a proglacial valley, Cordillera Blanca, Peru, Hydrol. Process., 30, 2915–2929, doi:10.1002/hyp.10912 (2016).
M. Hayashi, Alpine hydrogeology: The critical role of groundwater in sourcing the headwaters of the world, Groundwater, 58(4), 498–510, doi:10.1111/gwat.12965 (2020).
M. Huss, R. Hock, Global-scale hydrological response to future glacier mass loss, Nature Climate Change, 8, 135–140, doi:10.1038/s41558-017-0049-x (2018).
P. H. Gleick, M. Palaniappan, Peak water limits to freshwater withdrawal and use, PNAS, 107(25), 11155–11162, doi:10.1073/pnas.1004812107 (2010).
M. Baraer, B. Mark, J. McKenzie, T. Condom, J. Bury, K. Huh, C. Portocarrero, J. Gómez, S. Rathay, Glacier recession and water resources in Peru’s Cordillera Blanca. Journal of Glaciology, 58(207), 134–150. doi:10.3189/2012JoG11J186 (2012).
G. McDowell, M. Koppes, L. Harris, K. M. A. Chan, M. F. Price, D. G. Lama, G. Jiménez, Lived experiences of ‘peak water’ in the high mountains of Nepal and Peru, Climate and Development, doi:10.1080/17565529.2021.1913085 (2021).
A. S. Downing, E. H. van Nes, W. M. Mooij, M. Scheffer, The Resilience and Resistance of an Ecosystem to a Collapse of Diversity. PLoS ONE, 7(9), e46135, doi:10.1371/journal.pone.0046135 (2012).
R. J. DeRose, J. N. Long, Resistance and resilience: A conceptual framework for silviculture, For. Sci., 60(6), 1205–1212, doi:10.5849/forsci.13-507 (2014).
Z. P. Meyers, M. D. Frisbee, L. K. Rademacher, N. S. Stewart-Maddox, Old groundwater buffers the effects of a major drought in groundwater-dependent ecosystems of the eastern Sierra Nevada (CA), Environmental Research Letters, 16, 044044, doi:10.1088/1748-9326/abde5f (2021).
M. Baraer, J. McKenzie, B. G. Mark, R. Gordon, J. Bury, T. Condom, J. Gomez, S. Knox, S. K. Fortner, Contribution of groundwater to the outflow from ungauged glacierized catchments: a multi-site study in the tropical Cordillera Blanca, Peru, Hydrological Processes, 29, 2561–2581, doi:10.1002/hyp.10386 (2015).
L. W. Gelhar, J. L. Wilson, Ground-water quality monitoring, Ground Water, 12(6), 399–408 (1974).
A. D. Erskine, A. Papaioannou, The use of aquifer response rate in the assessment of groundwater resources, Journal of Hydrology, 202, 373–391 (1997).
J. S. Chen, C.-Y. Wang, Rising springs along the Silk Road, Geology, 37(3), 243–246, doi: 10.1130/G25472A.1 (2009).
M. Stoelzle, T. Schuetz, M. Weiler, K. Stahl, L. M. Tallaksen, Beyond binary baseflow separation: A delayed-flow index for multiple streamflow contributions, Hydrol. Earth Syst. Sci., 24, 849–867, doi:org/10.5194/hess-24-849-2020 (2020).
A. F. McClymont, M. Hayashi, L. R. Bentley, J. Liard, Locating and characterizing groundwater storage areas within an alpine watershed using time-lapse gravity, GPR and seismic refraction methods, Hydrol. Process., 26, 1792–1804, doi:10.1002/hyp.9316 (2012).
Institute of Hydrology, Low Flow Studies Report no.1, Research Report, Wallingford, UK: Institute of Hydrology, retrieved from http://nora.nerc.ac.uk/id/eprint/9093/ (1980).
R. A. Sloto, M. Y. Crouse, HYSEP: A computer program for streamflow hydrograph separation and analyses, U.S. Geological Survey, Water-Resources Investigations Report 96-4040 (1996).

There is NO Competing Interest.

SupplementaryMaterial.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Groundwater provides temporary resilience to alpine catchments in glacial retreat

Status:

Version 1

Abstract

Figures

Introduction

Controls On Differences Between Pre-peak Water And Post-peak Water States

Controls On The Hydraulic Response Time Of Alpine Glacial Catchments

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1