Partitioning the uncertainty of ensemble projections of global glacier mass change

(1)

Supporting Information for

Partitioning the Uncertainty of Ensemble Projections of Global

Glacier Mass Change

Ben Marzeion, Regine Hock, Brian Anderson, Andrew Bliss, Nicolas Champollion,

Koji Fujita, Matthias Huss, Walter Immerzeel, Philip Kraaijenbrink, Jan-Hendrik

Malles, Fabien Maussion, Valentina Radić, David R. Rounce, Akiko Sakai, Sarah

Shannon, Roderik van de Wal, and Harry Zekollari

Contents of this file

. Figures S to S

. Tables S and S

. References

Overview over supplementary figures

Figure

Variable

Scenario

S

specific mass balance

RCP.

S

RCP.

S

RCP.

S

RCP.

S

glacier area change relative to



RCP.

S

RCP.

S

RCP.

S

RCP.

S

rates of glacier mass loss

RCP.

S

RCP.

S

RCP.

S

RCP.

S

temporally accumulated glacier mass loss since



RCP.

S

RCP.

S

RCP.

S

RCP.

S

glacier mass relative to



RCP.

S

RCP.

S

RCP.

S

RCP.

S

absolute contribution of each source of uncertainty to the total

uncer-tainty of projected mass loss rates

n/a

S

relative contribution of each source of uncertainty to the total

uncer-tainty of projected mass loss rates

n/a

S

absolute contribution of each source of uncertainty to the total

uncer-tainty of projected mass loss accumulated since



n/a

Overview over supplementary tables

Table

Description

S

definition of years in each glacier model’s submitted annual volume and area

time-series

S

summary of bias correction, downscaling, and calibration and evaluation of

mod-eled mass balance for each glacier model

(2)

2020 2040 2060 2080 2100 -4000 -3500 -3000 -2500 -2000 -1500 -1000 -500 0 500 M/ t (kg m -2 yr -1 ) 2020 2060 2100 -4000 -3000 -2000 -1000 0 2020 2060 2100 -4000 -3000 -2000 -1000 0

Arctic Canada (North)

2020 2060 2100 -4000 -3000 -2000 -1000 0 Alaska 2020 2060 2100 -4000 -3000 -2000 -1000 0 M/ t (kg m -2 yr -1) Greenland 2020 2060 2100 -4000 -3000 -2000 -1000 0 Russian Arctic 2020 2060 2100 -4000 -3000 -2000 -1000 0

Arctic Canada (South)

2020 2060 2100 -4000 -3000 -2000 -1000 0 Svalbard 2020 2060 2100 -4000 -3000 -2000 -1000 0 M/ t (kg m -2 yr -1 ) Southern Andes 2020 2060 2100 -4000 -3000 -2000 -1000 0 Central Asia 2020 2060 2100 -4000 -3000 -2000 -1000 0 Iceland 2020 2060 2100 -4000 -3000 -2000 -1000 0

South Asia (West)

2020 2060 2100 -4000 -3000 -2000 -1000 0 M/ t (kg m -2 yr -1)

Western Canada and U.S.

2020 2060 2100 -4000 -3000 -2000 -1000 0

South Asia (East)

2020 2060 2100 -4000 -3000 -2000 -1000 0 Scandinavia 2020 2060 2100 -4000 -3000 -2000 -1000 0 North Asia 2020 2060 2100 year -4000 -3000 -2000 -1000 0 M/ t (kg m -2 yr -1) Central Europe 2020 2060 2100 year -4000 -3000 -2000 -1000 0 Low Latitudes 2020 2060 2100 year -4000 -3000 -2000 -1000 0 Caucasus 2020 2060 2100 year -4000 -3000 -2000 -1000 0 New Zealand AND2012 GLIMB GloGEM GloGEMflow JULES KRA2017 MAR2012 OGGM PyGEM RAD2014 WAL2001 median

Figure S Specific mass balance in kg m

−

₂

yr

−

1 as a function of time and glacier model, for

scenario RCP

.. Colored solid lines indicate the medians of each glacier model’s multi-GCM

projections. Global panel: dotted black line indicates average of all regional ensemble medians,

weighted by glacier area of the respective region. Solid black line indicates median of the

medians of each glacier model’s multi-GCM projections with global coverage. Regional panels:

black lines indicate median of the medians of each glacier model’s multi-GCM projections

covering that region.

(3)

2020 2040 2060 2080 2100 -4000 -3500 -3000 -2500 -2000 -1500 -1000 -500 0 500 M/ t (kg m -2 yr -1 ) 2020 2060 2100 -4000 -3000 -2000 -1000 0 2020 2060 2100 -4000 -3000 -2000 -1000 0

2020 2060 2100 -4000 -3000 -2000 -1000 0 Svalbard 2020 2060 2100 -4000 -3000 -2000 -1000 0 M/ t (kg m -2 yr -1) Southern Andes 2020 2060 2100 -4000 -3000 -2000 -1000 0 Central Asia 2020 2060 2100 -4000 -3000 -2000 -1000 0 Iceland 2020 2060 2100 -4000 -3000 -2000 -1000 0

South Asia (West)

2020 2060 2100 -4000 -3000 -2000 -1000 0 M/ t (kg m -2 yr -1)

2020 2060 2100 -4000 -3000 -2000 -1000 0

South Asia (East)

Figure S As Fig. S, but for scenario RCP..

(4)

2020 2040 2060 2080 2100 -4000 -3500 -3000 -2500 -2000 -1500 -1000 -500 0 500 M/ t (kg m -2 yr -1 ) 2020 2060 2100 -4000 -3000 -2000 -1000 0 2020 2060 2100 -4000 -3000 -2000 -1000 0

South Asia (West)

2020 2060 2100 -4000 -3000 -2000 -1000 0 M/ t (kg m -2 yr -1)

2020 2060 2100 -4000 -3000 -2000 -1000 0

South Asia (East)

Figure S As Fig. S, but for scenario RCP..

(5)

2020 2040 2060 2080 2100 -4000 -3500 -3000 -2500 -2000 -1500 -1000 -500 0 500 M/ t (kg m -2 yr -1 ) 2020 2060 2100 -4000 -3000 -2000 -1000 0 2020 2060 2100 -4000 -3000 -2000 -1000 0

South Asia (West)

2020 2060 2100 -4000 -3000 -2000 -1000 0 M/ t (kg m -2 yr -1)

2020 2060 2100 -4000 -3000 -2000 -1000 0

South Asia (East)

Figure S As Fig. S, but for scenario RCP..

(6)

2020 2040 2060 2080 2100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 area (rel. to 2015) 2020 2060 2100 0 0.2 0.4 0.6 0.8 2020 2060 2100 0 0.2 0.4 0.6 0.8

1 Arctic Canada (North)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Alaska 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 area (rel. to 2015) Greenland 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Russian Arctic 2020 2060 2100 0 0.2 0.4 0.6 0.8

1 Arctic Canada (South)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Svalbard 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 area (rel. to 2015) Southern Andes 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Central Asia 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Iceland 2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (West)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 area (rel. to 2015)

2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (East)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Scandinavia 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 North Asia 2020 2060 2100 year 0 0.2 0.4 0.6 0.8 1 area (rel. to 2015) Central Europe 2020 2060 2100 year 0 0.2 0.4 0.6 0.8 1 Low Latitudes 2020 2060 2100 year 0 0.2 0.4 0.6 0.8 1 Caucasus 2020 2060 2100 year 0 0.2 0.4 0.6 0.8 1 New Zealand AND2012 GLIMB GloGEM GloGEMflow JULES KRA2017 MAR2012 OGGM PyGEM RAD2014 WAL2001 median

Figure S Glacier area change relative to  as a function of time and glacier model, for

scenario RCP.. Colored solid lines indicate the median of each glacier model’s projections

forced by different GCMs. Black lines indicate median of all glacier model medians covering

that region.

(7)

2020 2040 2060 2080 2100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 area (rel. to 2015) 2020 2060 2100 0 0.2 0.4 0.6 0.8 2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (West)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 area (rel. to 2015)

2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (East)

Figure S As Fig. S, but for scenario RCP..

(8)

2020 2040 2060 2080 2100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 area (rel. to 2015) 2020 2060 2100 0 0.2 0.4 0.6 0.8 2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (West)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 area (rel. to 2015)

2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (East)

Figure S As Fig. S, but for scenario RCP..

(9)

2020 2040 2060 2080 2100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 area (rel. to 2015) 2020 2060 2100 0 0.2 0.4 0.6 0.8 2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (West)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 area (rel. to 2015)

2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (East)

Figure S As Fig. S, but for scenario RCP..

(10)

2020 2040 2060 2080 2100 0 0.5 1 1.5 2 2.5 3 M/ t (mm SLE yr -1 ) 2020 2060 2100 0 0.5 2020 2060 2100 0 0.5

2020 2060 2100 -0.2 0 0.2 0.4 0.6 Alaska 2020 2060 2100 -0.2 0 0.2 0.4 M/ t (mm SLE yr -1) Greenland 2020 2060 2100 0 0.2 0.4 0.6 Russian Arctic 2020 2060 2100 0 0.05 0.1 0.15 0.2

2020 2060 2100 0 0.1 0.2 0.3 0.4 Svalbard 2020 2060 2100 0 0.1 0.2 M/ t (mm SLE yr -1) Southern Andes 2020 2060 2100 0 0.05 0.1 0.15 0.2 Central Asia 2020 2060 2100 0 0.05 0.1 Iceland 2020 2060 2100 0 0.05 0.1

0.15 South Asia (West)

2020 2060 2100 0 0.05 0.1 M/ t (mm SLE yr -1)

2020 2060 2100 -0.02

0 0.02 0.04

0.06 South Asia (East)

2020 2060 2100 0 5 10 15 10 -3 _Scandinavia 2020 2060 2100 -5 0 5 10 15 10 -3 _{North Asia} 2020 2060 2100 year 0 5 10 15 M/ t (mm SLE yr -1) 10-3 Central Europe 2020 2060 2100 year -5 0 5 10 15 10 -3 _{Low Latitudes} 2020 2060 2100 year 0 5 10 10 -3 _Caucasus 2020 2060 2100 year -2 0 2 4 6 10 -3 _{New Zealand} AND2012 GLIMB GloGEM GloGEMflow JULES KRA2017 MAR2012 OGGM PyGEM RAD2014 WAL2001 median

Figure S Rates of glacier mass loss in mm SLE yr

−

₁

as a function of time and glacier model for

scenario RCP

.. Colored solid lines indicate median of each glacier model’s projections forced

by different GCMs. Global panel: dotted black line indicates sum of all regional ensemble

medians. Solid black line indicates median of the medians of all glacier models with global

coverage. Regional panels: black lines indicate median of all glacier model medians covering

that region.

(11)

2020 2040 2060 2080 2100 0 0.5 1 1.5 2 2.5 3 M/ t (mm SLE yr -1 ) 2020 2060 2100 0 0.5 2020 2060 2100 0 0.5

2020 2060 2100 0 0.05 0.1 M/ t (mm SLE yr -1)

2020 2060 2100 -0.02

0 0.02 0.04

Figure S As Fig. S, but for scenario RCP..

(12)

2020 2040 2060 2080 2100 0 0.5 1 1.5 2 2.5 3 M/ t (mm SLE yr -1 ) 2020 2060 2100 0 0.5 2020 2060 2100 0 0.5

2020 2060 2100 0 0.05 0.1 M/ t (mm SLE yr -1)

2020 2060 2100 -0.02

0 0.02 0.04

Figure S As Fig. S, but for scenario RCP..

(13)

2020 2040 2060 2080 2100 0 0.5 1 1.5 2 2.5 3 M/ t (mm SLE yr -1 ) 2020 2060 2100 0 0.5 2020 2060 2100 0 0.5

2020 2060 2100 0 0.05 0.1 M/ t (mm SLE yr -1)

2020 2060 2100 -0.02

0 0.02 0.04

Figure S As Fig. S, but for scenario RCP..

(14)

2020 2040 2060 2080 2100 0 50 100 150 200 250 M (mm SLE) 2020 2060 2100 0 10 20 30 2020 2060 2100 0 10 20 30

2020 2060 2100 0 10 20 30 Alaska 2020 2060 2100 0 5 10 15 20 25 M (mm SLE) Greenland 2020 2060 2100 0 10 20 30 Russian Arctic 2020 2060 2100 0 5 10

2020 2060 2100 0 5 10 15 20 Svalbard 2020 2060 2100 0 5 10 M (mm SLE) Southern Andes 2020 2060 2100 0 5 10 Central Asia 2020 2060 2100 0 2 4 6 Iceland 2020 2060 2100 0 2 4 6

8 South Asia (West)

2020 2060 2100 0 1 2 3 M (mm SLE)

2020 2060 2100 0

1 2

3 South Asia (East)

2020 2060 2100 0 0.2 0.4 0.6 Scandinavia 2020 2060 2100 0 0.1 0.2 0.3 0.4 North Asia 2020 2060 2100 year 0 0.1 0.2 0.3 0.4 M (mm SLE) Central Europe 2020 2060 2100 year 0 0.1 0.2 0.3 Low Latitudes 2020 2060 2100 year 0 0.05 0.1 0.15 Caucasus 2020 2060 2100 year 0 0.05 0.1 0.15 New Zealand AND2012 GLIMB GloGEM GloGEMflow JULES KRA2017 MAR2012 OGGM PyGEM RAD2014 WAL2001 median

Figure S Temporally accumulated glacier mass loss in mm SLE since  as a function of

time and glacier model for scenario RCP

.. Colored solid lines indicate median of each glacier

model’s projections forced by different GCMs. Global panel: dotted black line indicates sum of

all regional ensemble medians. Solid black line indicates median of the medians of all glacier

models with global coverage. Regional panels: black lines indicate median of all glacier model

medians covering that region.

(15)

2020 2040 2060 2080 2100 0 50 100 150 200 250 M (mm SLE) 2020 2060 2100 0 10 20 30 2020 2060 2100 0 10 20 30

8 South Asia (West)

2020 2060 2100 0 1 2 3 M (mm SLE)

2020 2060 2100 0

1 2

3 South Asia (East)

Figure S As Fig. S , but for scenario RCP..

(16)

2020 2040 2060 2080 2100 0 50 100 150 200 250 M (mm SLE) 2020 2060 2100 0 10 20 30 2020 2060 2100 0 10 20 30

8 South Asia (West)

2020 2060 2100 0 1 2 3 M (mm SLE)

2020 2060 2100 0

1 2

3 South Asia (East)

Figure S As Fig. S , but for scenario RCP..

(17)

2020 2040 2060 2080 2100 0 50 100 150 200 250 M (mm SLE) 2020 2060 2100 0 10 20 30 2020 2060 2100 0 10 20 30

8 South Asia (West)

2020 2060 2100 0 1 2 3 M (mm SLE)

2020 2060 2100 0

1 2

3 South Asia (East)

Figure S As Fig. S , but for scenario RCP..

(18)

2020 2040 2060 2080 2100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 M (rel. to 2015) 2020 2060 2100 0 0.2 0.4 0.6 0.8 2020 2060 2100 0 0.2 0.4 0.6 0.8

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Alaska 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 M (rel. to 2015) Greenland 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Russian Arctic 2020 2060 2100 0 0.2 0.4 0.6 0.8

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Svalbard 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 M (rel. to 2015) Southern Andes 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Central Asia 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Iceland 2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (West)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 M (rel. to 2015)

2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (East)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 Scandinavia 2020 2060 2100 0 0.2 0.4 0.6 0.8 1 North Asia 2020 2060 2100 year 0 0.2 0.4 0.6 0.8 1 M (rel. to 2015) Central Europe 2020 2060 2100 year 0 0.2 0.4 0.6 0.8 1 Low Latitudes 2020 2060 2100 year 0 0.2 0.4 0.6 0.8 1 Caucasus 2020 2060 2100 year 0 0.2 0.4 0.6 0.8 1 New Zealand AND2012 GLIMB GloGEM GloGEMflow JULES KRA2017 MAR2012 OGGM PyGEM RAD2014 WAL2001 median

Figure S Glacier mass relative to  as a function of time and glacier model for scenario

RCP

.. Colored solid lines indicate median of each glacier model’s projections forced by

different GCMs. Global panel: dotted black line indicates average of all regional ensemble

medians, weighted by glacier mass of the respective region. Solid black line indicates median

of the medians of all glacier models with global coverage. Regional panels: black lines indicate

median of all glacier model medians covering that region.

(19)

2020 2040 2060 2080 2100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 M (rel. to 2015) 2020 2060 2100 0 0.2 0.4 0.6 0.8 2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (West)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 M (rel. to 2015)

2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (East)

Figure S As Fig. S, but for scenario RCP..

(20)

2020 2040 2060 2080 2100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 M (rel. to 2015) 2020 2060 2100 0 0.2 0.4 0.6 0.8 2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (West)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 M (rel. to 2015)

2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (East)

Figure S As Fig. S, but for scenario RCP..

(21)

2020 2040 2060 2080 2100 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 M (rel. to 2015) 2020 2060 2100 0 0.2 0.4 0.6 0.8 2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (West)

2020 2060 2100 0 0.2 0.4 0.6 0.8 1 M (rel. to 2015)

2020 2060 2100 0 0.2 0.4 0.6 0.8

1 South Asia (East)

Figure S As Fig. S, but for scenario RCP..

(22)

2020 2040 2060 2080 2100 0 0.2 0.4 0.6 0.8 1 1.2 (mm SLE yr -1 ) 2020 2060 2100 0 0.2 2020 2060 2100 0 0.05 0.1 0.15 0.2

2020 2060 2100 0 0.05 0.1 0.15 0.2 Alaska 2020 2060 2100 0 0.05 0.1 0.15 (mm SLE yr -1) Greenland 2020 2060 2100 0 0.05 0.1 0.15 Russian Arctic 2020 2060 2100 0 0.05

0.1 Arctic Canada (South)

2020 2060 2100 0 0.05 0.1 0.15 Svalbard 2020 2060 2100 0 0.02 0.04 0.06 (mm SLE yr -1) Southern Andes 2020 2060 2100 0 0.02 0.04 0.06 Central Asia 2020 2060 2100 0 0.01 0.02 0.03 0.04 Iceland 2020 2060 2100 0 0.02 0.04

2020 2060 2100 0 0.01 0.02 0.03 0.04 (mm SLE yr -1)

2020 2060 2100 0 0.005 0.01 0.015 0.02

South Asia (East)

2020 2060 2100 0 2 4 6 10 -3 _Scandinavia 2020 2060 2100 0 1 2 3 4 10-3 North Asia 2020 2060 2100 year 0 2 4 6 (mm SLE yr -1) 10-3 Central Europe 2020 2060 2100 year 0 2 4 6 10 -3 _{Low Latitudes} 2020 2060 2100 year 0 1 2 3 10 -3 _Caucasus 2020 2060 2100 year 0 1 2 3 10 -3 _{New Zealand} ensemble tot glac GCM RCP nat

Figure S Contribution of each source of uncertainty (Equations -) to the total uncertainty

(σ

tot

, Equation

) of projected mass loss rates in mm SLE yr

−

1 . Also shown is the uncertainty

obtained directly from the entire ensemble without any decomposition (σ

ensemble

).

(23)

2020 2040 2060 2080 2100 0 10 20 30 40 50 60 70 80 90 100 (% of tot 2 ) 2020 2060 2100 0 50 2020 2060 2100 0 50

2020 2060 2100 0 50 100 Alaska 2020 2060 2100 0 50 100 (% of tot 2 ) Greenland 2020 2060 2100 0 50 100 Russian Arctic 2020 2060 2100 0 50

2020 2060 2100 0 50 100 Svalbard 2020 2060 2100 0 50 100 (% of tot 2 ) Southern Andes 2020 2060 2100 0 50 100 Central Asia 2020 2060 2100 0 50 100 Iceland 2020 2060 2100 0 50

100 South Asia (West)

2020 2060 2100 0 50 100 (% of tot 2 )

2020 2060 2100 0

50

100 South Asia (East)

2020 2060 2100 0 50 100 Scandinavia 2020 2060 2100 0 50 100 North Asia 2020 2060 2100 year 0 50 100 (% of tot 2 ) Central Europe 2020 2060 2100 year 0 50 100 Low Latitudes 2020 2060 2100 year 0 50 100 Caucasus 2020 2060 2100 year 0 50 100 New Zealand 2 glac 2 GCM 2 RCP 2 nat

Figure S Relative contribution of each source of uncertainty to the total uncertainty (σ

tot

,

Equation

) of projected mass loss rates in mm SLE yr

−

1 .

(24)

2020 2040 2060 2080 2100 0 10 20 30 40 50 60 70 80 (mm SLE) 2020 2060 2100 0 5 10 2020 2060 2100 0 5 10

2020 2060 2100 0 5 10 Alaska 2020 2060 2100 0 2 4 6 8 (mm SLE) Greenland 2020 2060 2100 0 5 10 Russian Arctic 2020 2060 2100 0 1 2 3 4

2020 2060 2100 0 2 4 6 Svalbard 2020 2060 2100 0 1 2 3 (mm SLE) Southern Andes 2020 2060 2100 0 1 2 3 Central Asia 2020 2060 2100 0 0.5 1 1.5 2 Iceland 2020 2060 2100 0 0.5 1 1.5 2

South Asia (West)

2020 2060 2100 0 0.2 0.4 0.6 (mm SLE)

2020 2060 2100 0

0.2 0.4 0.6

2020 2060 2100 0 0.05 0.1 0.15 Scandinavia 2020 2060 2100 0 0.05 0.1 0.15 North Asia 2020 2060 2100 year 0 0.02 0.04 0.06 (mm SLE) Central Europe 2020 2060 2100 year 0 0.05 0.1 Low Latitudes 2020 2060 2100 year 0 0.01 0.02 0.03 Caucasus 2020 2060 2100 year 0 0.02 0.04 0.06 New Zealand ensemble tot glac GCM RCP nat

Figure S Contribution of each source of uncertainty (Equations -) to the total uncertainty

(σ

tot

, Equation

) of projected mass loss accumulated since  in mm SLE. Also shown is the

uncertainty obtained directly from the entire ensemble without any decomposition (σ

ensemble

).

(25)

Table S Definition of years in each glacier model’s submitted annual volume and area time

series. Type of year indicates whether the mass changes were computed over calendar years or

mass-balance years spanning roughly the period between two consecutive mass minima in the

glaciers’ annual cycle. Period refers to the specific period over which the annual mass changes

were computed.



Model

Type of year

Period (northern hemisph.)

Period (southern hemisph.)

AND

mass-balance

n/a

 April -  March

GLIMB

mass-balance

 Oct. -  Sept.

GloGEM

mass-balance

 Oct. -  Sept.

 Apr. -  March

GloGEMflow

mass-balance

 Oct. -  Sept.

n/a

JULES

calendar

 Jan. -  Dec.

KRA

calendar

 Jan. -  Dec.

n/a

MAR

mass-balance

 Oct. -  Sept.

 Apr. -  March

OGGM

mass-balance

 Oct. -  Sept.

 Apr. -  March

PyGEM

mass-balance

 Oct. -  Sept.

n/a

RAD

mass-balance

 Sept. -  Aug. (> 75

◦

N);

 Oct. -  Sept. (< 75

◦

N)

 March. -  Feb. (< −75

◦

S);

 Apr. -  March (> 75

◦

S)

WAL



calendar

 Jan. -  Dec.



_{Each modeling group submitted the time series of annual glacier volume and area for the years}

_{ - }

(referring to the beginning of the year). However, the definition of a year differed between the different glacier

models. The submitted data from eight groups referred to mass-balance years, while the data from the other three

groups referred to calendar years. In addition, the definition of a mass-balance year differed between the models.

The discrepancy between calendar and mass-balance years was not corrected for. However, in Equation

, for each

model and region, ∆T was computed over the time period that corresponds to the model’s and region’s definition

of a year based on monthly data.

(26)

T

able

S

Summary

of

bias

correction

(removing

di

ff

erences

betw

een

GC

M

clima

tol

ogy

and

observ

ed

clima

tol

ogy),

downscaling

(accoun

ting

the

di

ff

erences

of

the

scales

of

GC

M

resol

ution

and

moun

tain

topogr

aph

y),

as

w

ell

as

calibr

ation

and

ev

al

ua

tion/v

alida

tion

proced

ures

of

modeled

mass

balance

for

each

glacier

model.

Bias

corrections

are

typicall

y

deriv

ed

from

com

parison

to

observ

ation-based

gridded

da

ta

sets,

while

further

adjustmen

ts

as

part

of

downscaling

are

often

deriv

ed

from

liter

ature-based

or

calibr

ated

model

par

ameters.

In

some

models,

bias

correction

and

downscaling

are

lum

ped

in

to

one

step.

W

e

do

not

incl

ude

inf

orma

tion

in

this

table

concerning

the

models’

trea

tmen

t

of

v

ariability

of

atmospheric

conditions

within

the

domain

of

an

individ

ual

glacier

.

AND  GLIMB Gl oGEM Gl oGEMfl ow JULES KRA  bias correction additiv e tem per a-ture bias correction and m ul tiplica tiv e precipita tion bias correction, applied mon thl y based on local tem per ature and precipita tion in terpola ted from sta tion da ta (a t .  ◦ resol ution) additiv e tem per a-ture bias correction and m ul tiplica tiv e seasonal tem per -ature am plitude correction based on ERA -In terim; see "downscaling" for pre-cipita tion trea tmen t additiv e tem per ature bias correction and m ul tiplica tiv e precip-ita tion bias correction; adjustmen t of in ter -ann ual v ariability for both, based on ERA -In terim additiv e tem per ature bias correction and m ul tiplica tiv e precip-ita tion bias correction; adjustmen t of in ter -ann ual v ariability for both, based on ERA -In terim additiv e tem per ature bias correction and m ul tiplica tiv e bias correction for pre-cipita tion, surf ace pressure, long and shortw av e radia tion and wind speed using WFDEI clima tol ogy (W eedon et al.;  ) additiv e tem per ature and precipita tion bias correction based on WFDEI clima tol ogy (W eedon et al.;  ) downscaling constan t tem per ature lapse ra te; precipita-tion correction factor constan t precipita tion correction factor spa tiall y and season-all y v arying tem per a-ture lapse ra te; precip-ita tion correction fac-tor spa tiall y and season-all y v arying tem per a-ture lapse ra te; precip-ita tion correction fac-tor constan t tem per ature lapse ra te, constan t scaling factor for wind speed; see "bias correction " for ad-justmen t of other v ariables constan t tem per ature lapse ra te; no fur -ther adjustmen t of precipita tion calibr ation individ ual glaciers are calibr ated to re-gional specific mass chang e  - (-.  m w .e. a − 1) from un published da ta regional calibr ation to regional mean specific mass chang e  - from Gardner et al. ( ) individ ual glaciers are calibr ated to re-gional specific mass chang e  - from Gardner et al. ( ); fron tal ab-la tion calibr ated to ma tch regional sum from v arious studies for individ ual glaciers, using glacier specific mass chang e from W G MS ( ), or rel ying on estima tes from nearby and similar ly -size glaciers regional calibr ation using elev ation-dependen t ann ual specific mass chang e from W G MS ( ) optimiza tion of mass balance gr adien t us-ing observ ed regional mass balances from v arious studies ev al ua tion/v alida tion for individ ual glaciers using direct ann ual mass balance obser -v ations from W G MS ( ) for individ ual glaciers using direct ann ual mass balance obser -v ations from W G MS ( ) for individ ual glaciers using direct seasonal mass balance observ a-tions (glacier -wide, el-ev ation bands, poin t balance) from W G MS ( ) for individ ual glaciers using direct ann ual mass balance obser -v ations from W G MS ( ) regional ev al ua tion using elev ation-dependen t win ter and summer specific mass balance observ ations from W G MS ( ) regional ev al ua tion using independen t remotel y -sensed length chang e, ELA and snowline da ta from Gardelle et al. ( ) and Scher ler et al. ( )



(27)

T

able

S

con

tin

ued

MAR  OGG M PyGEM RAD  W AL  bias correction additiv e tem per ature and precipita tion bias correction based on CRU CL .  clima tol-ogy additiv e tem per ature and precipita tion bias correction based on CRU CL .  clima-tol ogy; in ter ann ual tem per ature v ariabil-ity scaled to CRU TS additiv e tem per ature bias correction and m ul tiplica tiv e precip-ita tion bias correction, adjustmen t of in ter -ann ual v ariability for both, based on ERA -In terim additiv e tem per ature bias correction using ERA - , m ul tiplica-tiv e precipita tion bias correction based on Beck et al. ( ); adjustmen t of in ter -ann ual v ariability for both no bias correction, model uses onl tem per ature and pre-cipita tion anomalies downscaling spa tiall y v arying tem per ature lapse ra te; precipita tion correction factor constan t tem per ature lapse ra te and precip-ita tion correction fac-tor spa tiall y and tem po-rall y v arying tem per -ature lapse ra te; no further adjustmen t of precipita tion spa tiall y v arying tem per ature lapse ra te and precipita tion correction factor none calibr ation for individ ual glaciers, using direct mass bal-ance observ ations from W G MS ( ) for individ ual glaciers, using direct mass bal-ance observ ations from W G MS ( ) for individ ual glaciers, using g eodetic mass balance observ ations from Shean et al. ( ) for individ ual glaciers, using seasonal mass balance profiles from  glaciers from W G MS ( ), for re-gions using regionall y com piled glaciol ogical and g eodetic mass chang e from C ogley ( ) for  individ ual glaciers, using mass chang e observ ations from Zuo and Oer mans ( ) ev al ua tion/v alida tion for individ ual glaciers, using glaciol ogical mass balance obser -v ations from W G MS ( ), lea v e-one-glacier -out cross-v alida tion for individ ual glaciers, using glaciol ogical mass balance obser -v ations from W G MS ( ), lea v e-one-glacier -out cross-v alida tion for individ ual glaciers, using g eodetic and glaciol ogical measure-men ts from W G MS ( ) and equilib-rium line al titudes from Gardelle et al. ( ) for individ ual glaciers, using seasonal mass chang e measuremen ts from Dyurg erov ( ) par ameter sensitivity anal ysis, details Slang en and v an W al ( )



(28)