1

Seasonal glacier melt contribution to streamflow Neil Schaner Department of Civil and Environmental Engineering Box 352700 University of Washington Seattle, WA 98195 schanerneil@hydro.washington.edu Nathalie Voisin Pacific Northwest National Laboratories P.O. Box 999 Richland, WA 99352 nathalie.voisin@pnl.gov Dennis P. Lettenmaier* Department of Civil and Environmental Engineering Box 352700 University of Washington Seattle, WA 98195 dennisl@u.washington.edu

2

Ongoing and projected future changes in glacier extent and water storage globally

have lead to concerns about the implications for water supplies. However, the

current magnitude of glacier contributions to river runoff is not well known, nor is

the population at risk to future glacier changes. We estimate an upper bound on

glacier melt contribution to seasonal streamflow by computing the energy balance of

glaciers globally. Melt water quantities are computed as a fraction of total

streamflow simulated using a hydrology model and the melt fraction is tracked

down the stream network. In general, our estimates of the glacier melt contribution

to streamflow are lower than previously published values. Nonetheless, we find that

globally an estimated 225 (36) million people live in river basins where maximum

seasonal glacier melt contributes at least 10% (25%) of streamflow, mostly in the

High Asia region.

One sixth of the world’s population, and one quarter of its gross domestic product,

resides in areas that rely on snow or glacier melt for a majority of its water supply (1).

Glaciers contribute significantly to water resources, especially in the High Asia region,

which is the headwaters of many of that continent’s largest rivers (2). Immerziel et al. (3)

estimated that glacier melt contribution to five major Southeast Asian rivers contributes

significantly to the food supply of at least 60 million people. While the regional volume

of glacier contribution may not always be large, melt water is important locally (4).

Despite concerns about glacier extent changes and implications for water supplies (5), the

global contribution of glaciers to water supply is not well known (6).

3

!

Several previous studies have attempted to analyze the contribution of glacier melt to

water supply, mostly in India, China, Peru, and Canada (see Table 2 for a review). The

majority of those studies indirectly derive the snow and seasonal glacier melt contribution

to streamflow by subtracting other components from the water balance. Other

approaches include simulation of glacier melt directly (7) and streamflow isotope

analysis to infer glacier melt contribution (8,9). Armstrong (6) summarizes the current

state of understanding glacier contributions to water supply as, “[p]revious assessments

of the glacier melt impact on surface water supply have been primarily either highly

qualitative or local in scale, and in some cases, appear to be simply incorrect.” In this

paper, we attempt to estimate seasonal glacier melt contributions to runoff globally by

using a combination of remote sensing data and modeling.

We define glacier melt as runoff whose source is perennial snow, firn, or ice. We include

all ice caps (ice sheets covering less than 50,000 km2) and other permanent ice globally

outside of Antarctica and Greenland. Our effort is directed towards estimating the

present contribution of glaciers to water supply; we do not attempt to project future

changes. Nonetheless, our results arguably are important for understanding climate

change impacts, as they provide a basis for identifying regions that are at risk to potential

changes in glacier-derived runoff.

4

Methods

Determining the contribution of glacier melt to streamflow globally is complicated by the

absence of meteorological data for all but a very small number of the world’s glaciers.

We therefore attempted to bound the glacier contribution to streamflow using an energy

balance approach applied to known glacierized areas, estimated primarily using remote

sensing data. We assumed that all available energy over glacierized areas is converted to

melt. We then compared these estimates to model estimates of river runoff, from which

we derived a fractional contribution of glacier melt. We focused on the maximum

seasonal contribution, which generally occurs in summer when remaining seasonal snow

melt is low, glacier melt is high, and other non-glacier sources of runoff are low.

The energy available for melt is the sum of net radiation (Qnet, incoming (SWdown) minus

reflected (!SWdown) shortwave plus net longwave (LWnet)), sensible (Hsens), latent (Hlat),

and advected heat. Compared to the other components, advected heat is generally small

and is ignored (10).

We used the Variable Infiltration Capacity (VIC) hydrological model (11) applied

globally at one-quarter degree latitude by longitude to predict runoff from non-glacier

sources at each grid cell. We then computed the sum of glacier and non-glacier source

runoff accumulated downstream. The ratio of accumulated glacier source to total runoff

was then computed by month for the period 1998-2006. Stream networks were derived

using flow direction data from Wu et al. (12).

5

We estimated radiation components over glacierized areas using the Thornton and

Running (13) and Tennessee Valley Authority (14) algorithms for downward solar and

longwave radiation, respectively, as implemented by Maurer et al. (15). We estimated

emitted longwave radiation using an emissivity of 1.0 and an assumed surface

temperature of 0°C during the melt period.

Qnet is particularly sensitive to albedo, which can vary widely for glaciers. Albedo

depends on the microstructure of snow and ice, and is highest for new fallen snow, and

decreases as snow melts or is metamorphosed into glacier ice. It also has angular and

spectral dependencies, which we accounted for only indirectly using a single effective

albedo, an assumption that is common in snow models (16). Albedo estimates for

glaciers have been mostly derived from direct observations of downward and reflected

solar radiation (over small areas, which are essentially points in the context of our model

estimates) and satellite-based estimates. We compiled 100 glacier ice albedo estimates

from 15 publications, which ranged from 0.03 to 0.85 with a median of 0.35. The highest

values are for glaciers covered partially or fully with recently fallen snow; the lowest

values are from glaciers with substantial debris, black carbon, and/or biomass growth.

Given our interest in estimating an upper bound, we used the lower 25th percentile of the

100 values (0.24). We chose not to use the lowest value, which likely is associated with

debris covered glaciers to the extent that other thermal factors (notably, insulation by the

debris cover) reduce melt in ways not accounted for by our estimates. Supplementary

6

Note Glacier Albedo contains additional information regarding the glacier albedo

estimates.

We used the Global Land Ice Measurements from Space (GLIMS) (17) and the Digital

Chart of the World (DCW) (18) data sets to identify glacier area globally. The GLIMS

data have a spatial resolution of approximately 150 m, from which we computed the

glacierized fraction of each VIC quarter-degree grid cell. GLIMS is known to

underestimate the glacierized area globally (19). We therefore supplemented the GLIMS

data with DCW, essentially by creating the union of the two data sets for glacier area.

DCW data are derived from 1:1,000,000 scale maps. The merger of GLIMS with DCW

likely overestimates total glacier area because DCW is known to include some ephemeral

snowfields; which we accepted given our interest in estimating an upper bound to glacier-

derived river runoff. We then used the glacierized fraction of each VIC quarter-degree

grid cell to compute the maximum glacier melt as describe above, averaged for the period

1998-2006.

Streamflow for the entire domain was simulated with the VIC model, which estimates

runoff from melt of seasonal (ephemeral) snow cover and other sources (in particular,

rainfall-derived direct runoff and baseflow). A modification to VIC was implemented so

that its computation of runoff from the glacierized portions of each grid cell was

consistent with our maximized melt contributions outlined above. For snowmelt

computations, VIC uses an energy balance approach similar to the simplified approach

7

we used for glacier melt, but with an algorithm that accounts for refreezing of melt water

in the pack, the role of vegetation where present, and represents the snowpack as two

layers for purposes of thermal computations. VIC also computes albedo using a snow-

aging function based on elapsed time since snowfall and time of year. Details of the VIC

snow model are provided in Andreadis et al. (16). The VIC model was run from January

1998 to December 2006 at a 3-hourly time step, from which runoff was accumulated to

monthly averages. Supplementary Methods contains details of the VIC model

simulations.

Results

We first derived maps of the global area for which river runoff could be affected by

glacier melt, regardless of the amount. This mask (Figure 1) constitutes the domain for

our VIC model simulations. We then computed the ratio of accumulated glacier-derived

runoff to total runoff for each grid cell in the domain, and identified areas for which the

maximum monthly fraction of glacier-derived runoff was greater than 5, 10, 25, and 50

percent. These areas are shown in Figure 2. The largest inferred fractions are mostly for

July in the northern hemisphere and January in the southern hemisphere. The largest

areas identified are streams with headwaters in the Himalayan Mountains, coastal Alaska,

the southern Andes, Iceland, and the Alps. A number of smaller areas were also

identified.

8

To assess populations potentially at risk to changes in glacier runoff, we overlaid the

Gridded Population of the World 2010 data (20) on areas shown in Figure 2. The results

show that for the 5% threshold, the population potentially affected is about 450 million

(6.6% of global population). For 10%, 25%, and 50% thresholds the estimated

populations are 225 million (3.3%), 36 million (0.5%), and 2 million (<0.1%) people,

respectively. Table 1 shows the affected populations by region.

Discussion

Two points stand out from the results. First, while the total domain for which there is any

inferred signature of glacier melt is quite large (58,700,000 km2 or about 44% of total

global land area exclusive of Greenland and Antarctica), the area for which runoff

exceeds even the lowest threshold (5%) for the maximum melt month is much smaller,

6,630,000 km2 or about 5% of global land area. Second, many areas identified in Figure

2 have relatively small populations.

We compared our estimates of fractional contribution of glacier melt to other sources

where available (Table 2). Where referenced values are given for river basins that

constitute more than one VIC grid cell, the grid cell corresponding to the basin outlet was

used. As a result, comparisons are most relevant for point references. The references

that use water balance approaches derive glacier contribution indirectly by subtracting

9

estimated water balance components from observed flow. Some of the published values

are known to include seasonal snowmelt and are asterisked.

In addition to the comparisons in Table 2, we applied our melt estimates to published

values for two United States Geological Survey (USGS) Benchmark glaciers in Alaska,

Gulkana and Wolverine (21). Both glaciers have long-term records of mass balance,

precipitation, and streamflow discharge. Wolverine Glacier is located in a maritime

environment, whereas Gulkana Glacier is in a continental environment (22). Gulkana

glacier covers 19.6 km2 in a drainage basin of 31.6 km2 (stream gauge location about one

km downstream of the glacier terminus). Wolverine glacier is similar in size (16.8 km2)

and lies in a 24.6 km2 drainage basin, the gauge for which is about 150 m downstream of

its terminus.

From 1998 to 2006, the month of maximum potential melt, based on Qnet, for Gulkana

glacier is July, with an average recorded discharge of 10.4 m3/s. This compares well with

our predicted July average discharge of 9.7 m3/s. Maximum potential melt occurs in June

for Wolverine glacier with values of 6.3 m3/s (recorded) and 5.2 m3/s (predicted). Given

our simplifying assumptions, this agreement is encouraging.

Taken as a group our estimates of glacier-derived runoff from Table 2 are lower than

published values, and the differences are larger than for the well instrumented Gulkana

and Wolverine glaciers. Our energy balance model does not account for all of the

10

complex processes occurring within and upon glaciers and their drainage basins. We

discuss below the major simplifications and their implications.

In our model, all glaciers are uniform, flat slabs with full exposure to solar radiation and a

fixed, relatively low albedo. The assumption of a glacier surface (and implicitly,

subsurface) temperature of 0ºC during the melt period will result in an overestimation of

melt (23) because refreezing within the glacier pack is not represented (24). Also, as

glacier melt water moves downstream it is compared to total runoff that outside the

modeling environment may be diverted by water engineering works and agriculture (25).

Diversions reduce total runoff, increasing the glacier fraction contribution.

Because we do not represent storage of liquid water within the glacier, modeled melt

water enters the stream network too soon (26, 27). This could put melt water runoff out

of phase with non-glacier runoff simulated by the VIC model. The direction of this error

depends on the nature of non-glacier derived runoff. In basins where summers are

relatively dry, streamflow decreases throughout the summer, and failure to represent the

delay creates a negative bias in the glacier contribution estimate. On the other hand, for

river basins with summer monsoon runoff, late summer runoff may exceed that in early

summer, and the bias will be in the opposite direction. This is likely the case for many

Asian rivers. Complex local glacierized environments may introduce errors in turbulent

heat flux estimates. However, net radiation is the main contributor to melt in both high-

energy conditions (clear and warm) and low-energy conditions (cloudy and cool) (28,

11

29), and this contribution is generally small, especially since latent and sensible heat are

usually of opposite sign. Advected energy is not included in our energy balance model

but in most cases is likely small (10). We do not consider the slope and aspect of glaciers

in our computation of net solar radiation; hence for south facing glaciers (northern

hemisphere) we likely underestimate net radiation, and hence melt. However, these

effects are likely greatest for small headwater glaciers, and should average out for larger

and/or multiple glaciers.

Biases in VIC-simulated runoff are another error source. VIC-simulated streamflow

generally agrees well with observations when the model forcings (especially

precipitation) are well known (15). To reduce precipitation biases associated with

topographic complexities, we used rescaled precipitation (30). The direction of VIC

simulation errors varies from watershed to watershed, and can be considered essentially

random on a global basis.

The contribution of non-seasonal (i.e., permanent, or at least long-term with respect to

our observation period) loss of glacier ice is another potential source of error. Our

method computes total melt, whether or not the glacier is in balance. However, we do

assume that the glacier area is fixed through the period of record. For most glaciers

however, changes in area over the 9-year analysis period are small relative to the area at

the beginning of the period. For receding glaciers, our approach will bias our estimates

upward somewhat.

12

As the climate warms, our estimates of glacier contribution fractions may increase as

mass loss accelerates, but ultimately will be more than cancelled by loss of glacier area,

and ultimately will begin to recede. It is important to note, however, that reduction in the

fraction of glacier runoff does not imply reduction of runoff and streamflow, which

depends on characteristics of precipitation and snow accumulation and ablation on non-

glacier fed portions of the river basins in question. Although not explicitly examined in

our analysis, glacier-derived runoff generally is less variable on an interannual basis than

non-glacier-derived runoff, so regardless of the direction and magnitude of changes in

total runoff that might accompany glacier recession, variability is likely to increase.

Conclusions

Our analysis provides a worldwide upper-bound estimate of the glacier-melt contribution

to river flows and populations potentially at risk to glacier retreat. We find that at least

6.6% of the global population lives in river basins that depend on seasonal glacier melt

for at least 5% of runoff in the peak melt month; at least 3.3% rely on 10% of glacier melt

runoff in the peak-melt month, at least 0.5% rely on 25% and less than 0.1% rely on 50%.

In general, our estimates of the glacier melt contribution to streamflow are lower than

previously published values (7-9, 31-39). Nonetheless, most of our assumptions result in

an upward bias in our estimates, which we argue serve as plausible upper bounds on

glacier-derived streamflow globally.

13

References

1. Barnett, T. P., Adam, J. C., Lettenmaier, D. P. Potential impacts of a warming

climate on water availability in snow-dominated regions. Nature 438 (2005).

2. Hock, R. Glacier melt: a review of processes and their modeling. Prog. Phys. Geog.

29, 362-391 (2005).

3. Immerziel, W. W., van Beek, L. P. H., Bierkens, M. F. P. Climate change will affect

the Asian water towers. Science 328, 1382-1385 (2010).

4. Weingartner, R., Viviroli, D., Schaedler, B. Water resources in mountain regions: a

methodological approach to assess the water balance in a highland-lowland system.

Hydrol. Process. 21, 578-585 (2007).

5. Cruz, R.V., et al. Climate Change 2007: Impacts, Adaptation and Vulnerability.

Contribution of Working Group II to the Fourth Assessment Report of the

Intergovernmental Panel on Climate Change. Parry, M. L., Canziani, O. F.,

Palutikof, J. P., van der Linden, P. J., Hanson, C. E., Eds. (Cambridge Univ. Press,

2007).

6. Armstrong, R. L. The glaciers of the Hindu Kush-Himalayan region: A summary of

the science regarding glacier melt/retreat in the Himalayan, Hindu Kush,

Karakorum, Pamir, and Tien Shen mountain ranges. (International Centre for

Integrated Mountain Development, Kathmandu, 2010).

14

7. Zhang, Y., Liu, S., Xu, J., Shangguan, D. Glacier change and glacier runoff

variation in the Tuotuo River basin, the source region of the Yangtze River in

western China. Environ. Geol. 56, 59-68 (2007).

8. Mark, B. G., Seltzer, G. O. Tropical glacier meltwater contribution to stream

discharge: a case study in the Cordillera Blanca, Peru. J. Glaciol. 49, 165, 271-281

(2003).

9. Mark, B. G., McKenzie, J. M., Gómez, J. Hydrochemical evaluation of the changing

glacier meltwater contribution to stream discharge: Callejon de Huaylas, Peru.

Hydrolog. Sci. J. 50, 6, 975-985 (2005).

10. LaChapelle, E. Annual mass and energy exchange on the Blue Glacier. J. Geophys.

Res. 64, 4, 443-449 (1959).

11. Liang, X., Lettenmaier, D. P., Wood, E. F., Burges, S. J. A simple hydrologically

based model of land surface water and energy fluxes for GSMs. J. Geophys. Res.

99, 14,415-14,428 (1994).

12. Wu, H., Kimball, J. S., Mantua, N., Stanford, J. Automated upscaling of river

networks for macroscale hydrological modeling. Water Resour. Res. 47 (2011).

13. Thornton, P. E., Running, S. W. An improved algorithm for estimating incident

daily solar radiation from measurements of temperature, humidity and precipitation.

Agr. Forest. Meteorol. 93, 211-228 (1999).

14. Tennessee Valley Authority (TVA). Heat and mass transfer between a water surface

and the atmosphere. Water Resour. Res. (1972).

15

15. Maurer, E. P., Nijssen, B., Lettenmaier, D. P. Use of reanalysis land surface water

budget variables in hydrologic studies. GEWEX News 4, 6–8 (2000).

16. Andreadis, K., Storck, P., Lettenmaier, D. P. Modeling snow accumulation and

ablation processes in forested environments. Water Resour. Res. 45 (2009).

17. Raup, B. H., Kieffer, H. H., Hare, T. M., Kargel, J. S. Generation of Data

Acquisition Requests for the ASTER Satellite Instrument for Monitoring a Globally

Distributed Target: Glaciers. IEEE T. Geosci. Remote 38, 1105-1112 (2000).

18. Penn State University Libraries Digital Chart of the World Server. Digital Chart of

the World (DCW) [Data File]. (http://www.maproom.psu.edu/dcw/).

19. National Snow and Ice Data Center (NSIDC) Global Land Ice Measurements from

Space (GLIMS) Glacier Database. GLIMS Glacier Database Status.

(http://glims.colorado.edu/glacierdata/data_completeness_status.php, 2010).

20. Socioeconomic Data and Applications Center (SEDAC). Gridded Population of the

World, Version 3 (GPWv3). (http://sedac.ciesin.columbia.edu/gpw/index.jsp, 2005).

21. United States Geological Survey (USGS). Benchmark Glaciers [Data File].

(http://ak.water.usgs.gov/glaciology/index.html).

22. Bitz, C. M., Battisti, D. S. Interannual to decadal variability in climate and the

glacier mass balance in Washington, Western Canada, and Alaska. J. Climate 12,

3181-3196 (1999).

23. Pellicotti, F., Carenzo, M., Helbing, J., Rimkus, S., Burlando, P. On the role of

subsurface heat conduction in glacier energy-balance modeling. Ann. Glaciol. 50,

50, 16-24 (2009).

16

24. Pfeffer, W. T., Sassolas, C., Bahr, D. B., Meier, M. F. Response time of glaciers as

a function of size and mass balance: 2. Numerical experiments. J. Geophys. Res.

103, B5, 9783-9789 (1998).

25. Kaser, G., Grosshauser, M., Marzeion, B. Contribution potential of glaciers to water

availablility in different climate regimes. P. Natl. A. Sci. USA 107, 47, 20223-20227

(2010).

26. Stenborg, T. Delay of run-off from a glacier basin. Geografiska Annaler. Series A,

Physical Geography 52, 1, 1-30 (1970).

27. Tangborn, W. V., Krimmel, R. M., Meier, M. F. A comparison of glacier mass

balance by glaciological, hydrological and mapping methods, South Cascade

Glacier, Washington. IAHS-AISH Publ. No. 104 (1975).

28. Brock B. W., Willis, I. C., Sharp, M. J., Arnold, N. S. Modelling seasonal spatial

variations in the surface energy balance of Haut Glacier d’Arolla, Switzerland. Ann.

Glaciol. 31, 53-62 (2000).

29. Datt, P., Srivastava, P. K., Negi, P. S., Satyawali, P. K. Surface energy balance of

seasonal snow cover for snow-melt estimation in N-W Himalaya. J. Earth Syst. Sci.

117, 5, 567-573 (2008).

30. Sheffield, J., Goteti, G., Wood, E. F. Development of a 50-yr high-resolution global

dataset of meteorological forcings for land surface modeling. J. Clim. 19, 13, 3088-

3111 (2006).

31. Jain, C. K. A hydro-chemical study of a mountainous watershed: the Ganga, India.

Water Res. 36, 1262-1274 (2002).

17

32. Kumar, V., Singh, P., Singh, V. Snow and glacier melt contribution in the Beas

River at Pandoh Dam, Himachal Pradesh, India. Hydrolog. Sci. J. 52, 2, 376-388

(2007).

33. Singh, P., Jain, J. K., Kumar, N. Estimation of snow and glacier-melt contribution to

the Chenab River, Western Himalaya. Mt. Res. Dev. 17, 1, 49-56 (1997).

34. Singh, P., Jain, S. K. Snow and glacier melt in the Satluj River at Bhakra Dam in the

western Himalayan region. Hydrolog. Sci. J. 47, 1, 93-106 (2002).

35. Wang, Z. Recent Variations of Glaciers and the Influence of Climate and Glacier

Fluctutaions on Runoff in China. In: Snow Cover and Glacier Variations. IAHS

Publ. No. 183 (1989).

36. Xu, B., Yao, T., Wang, N. Observed Glacier Melting and Black Soot Deposition in

the Third Pole, paper presented at Workshop on Ice, Snow and Water: Impacts of

Climate Change on California and Himalayan Asia, La Jolla, CA, 4 May 2009.

37. Yang, Z. An analysis of the water balance and water resources in the mountainous

Heihe River Basin. IAHS Publ. No. 183 (1989).

38. Comeau, L. E. L., Pietroniro, A., Demuth, M. N. Glacier contribution to the North

and South Saskatchewan Rivers. Hydrol. Process. 23, 2640-2653 (2009).

39. Hopkison, C., Young, G. J. The effect of glacier wastage on the flow of the Bow

River at Banff, Alberta, 1951-1993. Hydrol. Process. 12, 1745-1762 (1998).

Supplementary Information is linked to the online version of the paper at

www.nature.com/nclimate.

18

Acknowledgements

The research on which this paper is based was supported in part by Grant No.

NNX10AP90G from the National Aeronautics and Space Administration to the

University of Washington.

The data reported in this paper are archived at

ftp://ftp.hydro.washington.edu/pub/HYDRO/data/published/Schaneretal_2011_Global/

Author Contributions

D.P.L. conceived and supervised the project. N.S. and N.V. assembled input data, ran the

model, and analyzed the output data. N.S. and D.P.L. drafted the manuscript. All

authors commented on the manuscript at all stages.

Competing Financial Interests

The authors declare no competing financial interests.

Figure Legends

Figure 1: VIC model simulation domain

Figure 2: River basins for which at least 5% (green), 10% (yellow), 25% (orange), 50%

(red) of discharge is derived from glaciers in at least one month.

19

Tables Table 1: Estimates of affected populations and land areas

Threshold Contribution (%) 5 10 25 50

Total 447 225 36 2.2

Asia 424 216 34 2.1

North America 5.6 1.6 0.3 < 0.1

North America excluding Alaska 5.4 1.4 0.2 < 0.1

South America 4.5 2.9 0.7 < 0.1

Popu

latio

n A

ffect

ed

(mill

ions

of p

erso

ns)

Europe 13.4 5.1 0.2 -

Total (excluding Greenland and Antarctica)

5.0 3.0 0.9 0.2

Asia 7.1 4.2 1.2 0.2

North America 5.7 4.1 1.7 0.6

North America excluding Alaska 2.4 1.2 0.5 0.2

South America 5.1 2.1 0.8 0.2

Perc

enta

ge o

f Lan

d A

rea

(%)

Europe 5.2 0.5 0.0 -

20

Table 2: Comparison of published and calculated glacier contribution values

Source Area/River Method

Referenced glacier

contribution (%)

Calculated upper bound glacier

contribution (our analysis) (%)

Jain (30) Deoprayag, Ganga River, India - 28.7* 14.7

Kumar et al. (31)

Pandoh Dam, Beas River, India Water balance 37.4*

(1998-2004) 21.7

Singh et al. (32)

Akhnoor, Chenab River, India Water balance 49.1*

(1982 - 1992) 32.5

Singh and Jain (33)

Bhakra Dam, Satluj River, India Water balance 59*

(1986 –1996) 5.5

Wang (34) Muzat, China Mass and water balance 82.8 7.3

Tarim Basin, China - 40.2 24.2 Junggar Basin, China - 13.5 6.1 Qaidam Basin, China - 12.5 8.7 Hexi Corridor, China - 13.8 1.6

Xu et al. (35)

Qinghai Lake. China - 0.4 9.0

Yang (36) Heihe (Yingluxia Hydro Station), China Water balance 5 2.5

Zhang et al. (5) Tuotuo River, China Modified degree

day model 32

(1961-2004) 16.9

Yanamarey, Cordillera Blanca, Peru Water balance 35 ± 10

(1998-1999) 3.9

Uruashraju, Cordillera Blanca, Peru Water balance 36 ± 10

(1998-1999) 3.9 Mark and Seltzer (6)

Río Santa, Callejon de Huaylas, Peru

Hydrochemical mixing model

12-20 (1998-1999) 3.9

Yanamarey, Cordillera Blanca, Peru Water balance 58 ± 10

(2001-2004) 3.9 Mark et al. (7) Río Santa, Callejon de

Huaylas, Peru Hydrochemical mixing model

40 (2001-2004) 3.9

North Saskatchewan River at Edmonton, Alberta, Canada

WATFLOOD hydrological

model

9.33 (1975-1998) 6.1

Comeau et al. (37) Bow River at Calgary,

Alberta, Canada

WATFLOOD hydrological

model

6.25 (1975-1998) 5.0

Hopkinson and Young (38)

Bow River, Banff, Alberta, Canada

Mass and water balance 1.8-56** 31.4

* Referenced Glacier Contribution values are stated to include ephemeral snowmelt. **Average 1952-1993 and August 1970, respectively.

Supplementary Note: Glacier Albedo

Glacier energy balance is particularly sensitive to albedo. Albedo is highly

variability over small areas and interannually (1-3). Our global calculations for

determining a maximum threshold of glacier melt use a universal albedo value. From 15

publications, 100 glacier ice albedo values were extracted that range from 0.03 to 0.85 (1-

15). High values are almost certainly for cases where glacier ice is partially or fully

covered by recently fallen snow. The lowest values are from studies of glaciers that are

substantially covered by debris, black carbon, and/or biomass growth. The values fall

into three categories: those stated in the text or in a table in published articles or reports,

those derived from figures, and those averaged from a stated minimum and maximum. In

the case of continuous time series of albedo values, the average value was taken. The

statistics of the resulting 100 values are: mean = 0.39, standard deviation = 0.20, and

median = 0.35. Given our interest in estimating an upper bound, we used the lower 25th

percentile of the 100 values, 0.24. We chose not to use the lowest value, which is likely

associated with debris cover to the extent that other thermal factors (notably, insulation

by the debris cover) reduce melt in ways not accounted for by our estimates.

References

1. Bahadur, J. Himalayan Snow and Glaciers, Associated Environmental Problems,

Progress and Prospects. (Concept Publishing Company, New Delhi, India, 2004).

2. Hock, R., Holmgren, B. A distributed surface energy-balance model for complex

topography and its application to Storglaciären, Sweden. J. Glaciol. 51, 172 (2005).

3. Ohmura, A. Physical Basis for the Temperature-Based Melt-Index Method. J. Appl.

Meteorol. 40, 753-761 (2001).

4. Oerlemans, J., Giesen, R. H., van den Broeke, M. R. Retreating alpine glaciers:

increased melt rates due to accumulation of dust (Vadret da Morteratsch,

Switzerland). J. Glaciol. 55, 192, 729-736 (2009).

5. Greuell, W., et al. Assessment of interannual variations in the surface mass balance

of 18 Svalbard glaciers from the Moderate Resolution Imaging

Spectroradiometer/Terra albedo product. J. Geophys. Res., 112 (2007).

6. Brock, B. W., Willis, I. C., Sharp, M. J. Measurement and parameterization of

albedo variations at Haut Glacier d’Arolla, Switzerland. J. Glaciol. 46, 675-688

(2000).

7. Corripio, J. G. Snow surface albedo estimation using terrestrial photography. Int. J.

Remote Sens. 25, 24, 5705-5729 (2004).

8. Mattson, L. E. The influence of a debris cover on the mid-summer discharge of

Dome Glacier, Canadian Rocky Mountains in Debris-Covered Glaciers

(International Association of Hydrological Sciences, Publ. No. 264, 2000).

9. Weihs, P., et al. Modeling the effect of an inhomogeneous surface albedo on

incident UV radiation in mountainous terrain: determination of an effective surface

albedo. Geophys. Res. Lett. 28, 16, 3111-3114 (2001).

10. Paul, F., Haeberli, W. Spatial variability of glacier elevation changes in the Swiss

Alps obtained from two digital elevation models. Geophys. Res. Lett. 35 (2008).

11. Andreassen, L. M., van den Broeke, M. R., Giesen, R. H., Oerlemans, J. A 5 year

record of surface energy and mass balance from the ablation zone of Storbreen,

Norway. J. Glaciol. 54, 185, 245-258 (2008).

12. Fischer, A. Glacier and climate change: interpretation of 50 years of direct mass

balance of Hintereisferner. Global Planet. Change 71, 13-26 (2010).

13. Pellicciotti, F., et al. A study of the energy balance and melt regime on Juncal Norte

Glacier, semi-arid Andes of central Chile, using melt models of different complexity.

Hydrol. Process. 22, 3980–3997 (2008).

14. Sicart, J. E., Hock, R., Six, D. Glacier melt, air temperature, and energy balance in

different climates: the Bolivian Tropics, the French Alps, and northern Sweden. J.

Geophys. Res. 113, D24113 (2008).

15. Takeuchi, N. Temporal and spatial variations in spectral reflectance and

characteristics of surface dust on Gulkana Glacier, Alaska Range. J. Glaciol. 55,

192 (2009).

Supplementary Methods

The Variable Infiltration Capacity (VIC) Model (1) was used to simulate global

energy and water balance components. The VIC model is a semi-distributed macro scale

hydrology model. The model was run at one-quarter degree latitude by longitude spatial

resolution for all drainage basins with any glacier runoff contribution (Figure 1).

Daily precipitation, minimum and maximum temperatures, and wind speed are the

primary forcings for the VIC model (other forcings, such as downward solar and

longwave radiation, and humidity, are derived from these variables). Precipitation data

were taken from the Tropical Rainfall Measurement Mission (TRMM) Multi-satellite

Precipitation Analysis (TMPA) for 50ºS – 50ºN latitude band (2). Outside of this band,

precipitation was taken from Sheffield et al. (3). These data, which have a spatial

resolution of one-degree latitude by longitude, were interpolated to quarter-degree using

the inverse distance squared SYMAP algorithm (4, 5). Temperature and wind data were

taken from the National Centers for Environmental Prediction (NCEP) Reanalysis 2 data

set (6). Wind was interpolated from the Gaussian grid to the regular quarter degree grid

using the SYMAP algorithm. The daily temperature minima and maxima were

interpolated and adjusted using a pseudo-adiabatic lapse rate of 0.065º/1000m using the

SYMAP algorithm and the Global Land One-kilometer Base Elevation Project (GLOBE)

digital elevation model (DEM) (7).

Global soil parameters were derived as described in Nijssen et al. (8) using the

2009 Harmonized World Soil Database (9) and remaining structure-based parameters

from Cosby et al. (10). Soil depths and baseflow parameters were taken from Nijssen et

al. (11). Global vegetation parameters were derived as in Su et al. (12). A maximum of

five VIC elevation bands (sub grid partitioning based on elevation for more accurate

snow-related simulations) per grid cell were created with the GLOBE DEM. Radiation

components over glacierized areas were estimated using the Thornton and Running (13)

and Tennessee Valley Authority (14) algorithms for downward solar and longwave

radiation, respectively. We estimated emitted longwave radiation using an emissivity of

1.0 and the VIC surface temperature, which during the melt period was assumed to be

0°C. For purposes of representing stream networks, we used the flow direction file of

Wu et al. (15).

The VIC model was initially run at a daily time step for a total of 18 years to

initialize the model’s soil water storages, and to create a snow “reservoir” in the areas

defined as glaciers. Following the 18-year spin-up the model was run at a 3-hourly time

step for 1998-2006.

The VIC model does not represent glaciers explicitly, and therefore treats glacier

areas as seasonally ephemeral snowpack within an energy balance snow accumulation

and ablation model, details of which are included in Andreadis et al. (16). To more

accurately represent glaciers within VIC, a snow “reservoir,” or modified snow pack, was

created for each gridcell containing glaciers. The modified snow pack representing

glaciers was made permanent with an artificially high value of snow water equivalent.

Albedo of the modified snowpack follows the default VIC calculation for snow albedo,

U.S. Army Corps of Engineers empirical snow albedo decay curves (17). During the

melt season, with little snowfall, the modified snowpack surface albedo decays to values

equivalent to known glacier ice albedo values. The modified snowpack is located in the

highest elevation bands (mountain glaciers) unless the total elevation difference within

the gridcell is less than 200 meters (valley or tidewater glaciers). The modified

snowpack within each glacierized gridcell covers each elevation band either wholly or is

absent. Therefore the areal coverage of the modified snowpack is greater than or equal to

the glacier area determined with the Global Land Ice Measurements from Space

(GLIMS) and Digital Chart of the World (DCW) datasets (18, 19).

References

1. Liang, X., Lettenmaier, D. P., Wood, E. F., Burges, S. J. A simple hydrologically

based model of land surface water and energy fluxes for GSMs. J. Geophys. Res.

99, 14,415-14,428 (1994).

2. Huffman, G. J., et al. The TRMM Multisatellite Precipitation Analysis (TMPA):

Quasi-Global, Multiyear, Combined-Sensor Precipitation Estimates at Fine Scales.

J. Hydrometeor. 8, 38–55 (2007).

3. Sheffield, J., Goteti, G., Wood, E. F. Development of a 50-yr high-resolution global

dataset of meteorological forcings for land surface modeling. J. Clim. 19, 13, 3088-

3111 (2006).

4. Shepard, D. S. Computer mapping: the SYMAP interpolation algorithm in Spatial

Statistics and Models, Gaile, Willmott, eds. (1984).

5. Huffman, G. J., et al. Global precipitation at one-degree daily resolution from multi-

satellite observations. J. Hydromet. 2, 36-50 (2001).

6. Kanamitsu, M., et al. NCEP-DEO AMIP-II Reanalysis (R-2). Bul. of the

Atmos. Met. Soc. 1631-1643 (2002).

7. Hastings, D. A., et al. The Global Land One-kilometer Base Elevation (GLOBE)

Digital Elevation Model, Version 1.0 [Data File]. (National Oceanic and

Atmospheric Administration, National Geophysical Data Center,

http://www.ngdc.noaa.gov/mgg/topo/globe.html, 1999).

8. Nijssen, B., Schnur, R., Lettenmaier, D. P. Global retrospective estimation of soil

moisture using the Variable Infiltration Capacity land surface model, 1980-1993. J.

Clim. 14, 1790-1808 (2001).

9. FAO/IIASA/ISRIC/ISSCAS/JRC, Harmonized World Soil Database (version 1.1)

(FAO, Rome, Italy and IIASA, Laxenburg, Austria, 2009).

10. Cosby, B. J., Hornberger, G. M., Clapp, R. B., Ginn, T. R. A statistical exploration

of the relationships of soil moisture characteristics to the physical properties of soils.

Water Resour. Res. 20, 682–690 (1984).

11. Nijssen, B., O'Donnell, G. M., Lettenmaier, D. P., Lohmann, D., Wood, E. F.

Predicting the discharge of global rivers. J. Clim. 14, 3307-3323 (2001).

12. Su, F., Adam, J. C., Bowling, L. C., Lettenmaier, D. P. Streamflow Simulations of

the Terrestrial Arctic Domain , J. Geophys. Res. 110 (2005).

13. Thornton, P. E., Running, S. W. An improved algorithm for estimating incident

daily solar radiation from measurements of temperature, humidity and precipitation.

Agr. Forest. Meteorol. 93, 211-228 (1999).

14. Tennessee Valley Authority (TVA). Heat and mass transfer between a water surface

and the atmosphere. Water Resour. Res. (1972).

15. Wu, H., Kimball, J. S., Mantua, N., Stanford, J. Automated upscaling of river

networks for macroscale hydrological modeling. Water Resour. Res. 47 (2011).

16. Andreadis, K., Storck, P., Lettenmaier, D. P. Modeling snow accumulation and

ablation processes in forested environments. Water Resour. Res. 45 (2009).

17. U.S. Army Corps of Engineers North Pacific Division. Snow hydrology: summary

report of the snow investigations. (1956) pp. 437.

18. Penn State University Libraries Digital Chart of the World Server. Digital Chart of

the World (DCW) [Data File]. (http://www.maproom.psu.edu/dcw/).

19. Raup, B. H., Kieffer, H. H., Hare, T. M., Kargel, J. S. Generation of Data

Acquisition Requests for the ASTER Satellite Instrument for Monitoring a Globally

Distributed Target: Glaciers. IEEE T. Geosci. Remote 38, 1105-1112 (2000).

Figure 1: VIC model simulation domain

Figure 2: River basins for which at least 5% (green), 10% (yellow), 25% (orange), 50% (red) of discharge is derived from glaciers in at least one month.