Challenges of Modelling Surface Mass Balance

9/15/16

Advanced Training Course on Remote Sensing of the Cryosphere Leeds (UK), 12-16 September 2016

Challenges of modelling surface mass balance

Michiel van den Broeke Institute for Marine and Atmospheric Research, Utrecht University (IMAU)

How do we define surface mass balance (SMB)?

Here: the sum of surface and internal mass balance (climatic mass balance or firn mass balance)

Ligtenberg, PhD thesis, 2014

1 9/15/16

Mass Balance = Surface Mass Balance – Discharge MB = dM/dt = SMB – D

SMB is challenging: not one, but three balances!

Ice sheet mass balance (MB) MB = Surface mass balance – Discharge [Gt yr-1]

Surface mass balance (SMB) SMB = Precipitation – Sublimation – Runoff - Erosion [Gt yr-1]

Liquid water balance (LWB) Runoff = Rain + Condensation + Melt – Refreezing – Retention [Gt yr-1]

Surface energy balance (SEB) -2 M = SWnet + LW net + H + L + Gs [W m ]

J. Paul Getty Museum

2 !"#$"#%

!*,2.&%) OVC).&)(.>/)*2(%04#>.*&() #&-):35),*-%$$.&'

N%,*>%)(%&(.&' W.0&),*-%$ E&)(.>/)*2(%04#>.*&(

N7-G&-$%*" =-G%E+-$%*" >1-$%-GO$'01*+-G( 'P$'"#%*" !$.,#>%),*-%$

GNX!O),#(()>0%&-()9UYYZIUYJU;

=*&+$'#8Q(R'+$(S*&$'+#

' !"#$"#%

JZ)H%#0()*+).1%)("%%>),#(()1"#&'%()+0*,)GNX!O

!"#$% &'()*+,,%-./01' !"#$%&'()*+,,%+'-%.2'(**%-./01'

/"$-+3$%3(%3'(#2''$ M+''"G-".(%3'(#2''$

T'G%3*<"- -".(*$2'+#5(BFAU

["%)\(.,M$%0]) 1#(%=) X&>#01>.1#

V/>/O,WI!>

( !"#$"#%

X--%-)4#$/%)*+):35),*-%$$.&'=)#)1#(%)(>/-H)+*0)X&>#01>.1#

3#(()1"#&'%().&)UYY^) +0*,)GNX!O

,-+$%"(X*+Y-$2

X--%-)4#$/%)*+):35),*-%$$.&'=)#)1#(%)(>/-H)+*0)X&>#01>.1#

O$%4#>.*&)1"#&'%().&)UYY^) +0*,)O&4.(#>

,-+$%"(X*+Y-$2

$ Future SMB of Antarctica

is forced using fields of temperature, specific humidity, trade-off between computational expense and spatial detail; zonal and meridional wind components, and surface pres- doubling the grid resolution would multiply the computa- sure from either GCM or re-analysis output. Relaxation of tional time by a factor 10. Moreover, the annual integrated RACMO2 prognostic variables towards external forcings is SMB of the AIS at 55 km resolution (Van de Berg et al. restricted to the boundary relaxation zone (Fig. 1). External 2006) is similar to that at 27 km resolution (Lenaerts et al. forcings are updated every six hours and linearly interpo- 2012a). For the scenario runs, the largest uncertainty there- lated in time to yield accurate values in between. Sea fore derives not from the model resolution but from the surface temperatures and sea-ice extent are also prescribed chosen forcing model and scenario. Given this information, from the forcing model. The version of RACMO2 used for and the fact that a 27 km resolution run is ten times as this study includes a snow model that calculates tempera- expensive as a 27 km run, we chose 55 km as final resolu- ture, density and meltwater processes (percolation, retention. The model topography, grid resolution and lateral tion, refreezing and runoff) in the snow (Ettema et al. relaxation boundary of the domain are shown in Fig. 1. 2009), and an improved albedo scheme, where the snow For the period 1980–1999, a RACMO2 reference sim- albedo depends on snow grain size (Kuipers Munneke ulation, forced by ERA-40 re-analysis data from the et al. 2011). For this study, contributions from drifting European Centre for Medium-Range Weather Forecasts snow processes have not been included, because the (Uppala et al. 2005), was performed in order to check the module of Lenaerts and Van den Broeke (2012) was not yet reliability of the GCM-forced RACMO2 simulations. In !"#$"#% fully implemented when we started the simulations. this paper, ERA-40 has been used as forcing instead of its For contemporary climate studies of the AIS (1–30 years), successor ERA-Interim (Dee and et al. 2011), since the RACMO2 has been run on grids with 27 and 5.5 km hori- latter only covered the period 1989–2009 at the time the zontal resolution (Lenaerts et al. 2012a, b). However, for the RACMO2 simulations were started. Other RACMO2 number of simulation years considered here (660 years in simulations forced by re-analysis data (ERA-40 or ERA- total), a horizontal resolution of 55 km is considered a good Interim) yielded realistic SMB results over Antarctica

Fig. 1 Map of Antarctica showingN%'.*&#$) the model domain, the boundary relaxation zone (dotted1$.,#>%),*-%$) area) and model topography in meters above sea level -*,#.&

U_)6,)0%(*$/>.*&

123

3*-%$$%- ZI"*/0$H) M0%1.M.>#>.*&

N/&=)NX!3VU`ZBXa[ N%(*$/>.*&=)U_)6, =*&+$'#8Q(,'G32%*+(7-"(S'##'0

% !"#$"#%

Ekstrom & Automatic Weather Stations Ekstrom Shelf Ice !(!( Halvfarryggen Atka & !( !( Antarctica - 2016 Atka Bay !( Troll Soerasen Camp Maudheimvida !( !( Molodeznaya Joinville Is Nordenskiold Utsteinen !( !( Gerlache Strait !( Kohnen Base Neko Harbor Mizuho AWS 17 *# Cierva Cove AWS 14 Larsen Ice Shelf !(!( !. Halley V JASE2007*# Arelis Amsler Island *#*# *# Relay Station Bonaparte Pt *#!. !( !( AWS 18 Windy *# !( AWS 1 Inset Map A Hugo Is !( !. *#!( Dome Fuji (JPN) Bisco Island !( Butler Is Dome Fuji Whitlock *# *# Dismal Is !( Site 8 Filchner !( !( !( *# Baldrick AGO-3 Fossil Bluff Limbert $+ !( Druznaja-4 Whoop !( Korff Ice Rise Whoop China !( PANDA-North !( Lucia !( Sky-Blu !( Manuela $+AGO-2 Eagle !( Virginia *#!( Dome A PANDA-South*# !( Mt Brown Eneide !(!( "/ Union Glacier !( Rita !( Maria Austin*# *# Henry "/ Thiel Mountains *# Jang Bogo Zoraida Nico AGO-4 !( !( *# Penguin !( Kathie# Kominko- Thurston Island * Lola *# *# Slade (WAIS) *#Theresa Evans Knoll *# Erin AGO-1 !( *# Harry*# $+ Lindsey Island Alessandra Byrd *# Sabrina !( *# *# Snyder Bear Peninsula *# Elizabeth *# Emma Concordia Haupt *# AGO-5 Nunatak Rocks Janet *# Siple Dome *# Lettau Elaine $+ Dome C II !( *# Skinner Saddle Dome C *#!(!( DC N Casey Schwerdtfeger !( Marilyn !( !(!(!( Skiway Silvia!( Dome C (FRA) !( 0 50 km *# *# !( Darwin Gl !( 0 500 km Margaret Gill*# *# Central Valley Law Dome Summit/A Cape !( *#*#!( Poinsett Russkaya Vito*#*#*#*#$1*#$1!(!(!.!. *#$1$1*#$1*#$1$1$1*#!(.$1!.(!.(!!( !( Giulia Wilkins Runway *# *# !( West 4m / East ICAO *# Alexander Inset Map B See Inset Map B *# D-85 *# !( Modesta Tall Tower *#!( !(!( !( Sofia-B !( !( Irene Minna Bluff See Inset Map A !( !( *# !( Paola D-47*# !( Minna Bluff *# D-0 Linda Cape Hallett *# *#!( *#*# D-3 # Herbie Alley N Miers Valley Possession Is D-10 Emilia * !( Port Martin D-17 White Is !( Leningradskaya Garwood Valley !. Cp Denison *# Cp Spencer !( Lorne*# $1 !( Bratina Island $1 *# Pegasus N !. White Is S *# !.!. LTER AWS (21) lf $1 $1 Jules United States AWS International AWS he White Out !.!. S *#Willie Field $1 !. !. Ice Biesiada Crevasse $1 $1 $1 !.!. Mt Fleming *# Univ. of Wisconsin (UW) !( Australia !( Netherlands s $1 Ford Rock !.!. os N Crevasse Butter Pt - !( *# UW / Australia !( Belgium/Netherlands !( New Zealand R Ferrell *# $1 Tent Is !. !. Bull Pass *# Ferrar *# !( *# UW / China !( Brazil !( Norway $1 *!($1 !( Victoria Valley $1 Cp Royds !( *# UW / France !( China !( Russia Windless Cones Marble Pt II Victoria Lower Gl *# UW / Japan !( China/Australia !( South Korea Bight Laurie II Marble Pt *# UW / New Zealand !( Finland !( United Kingdom *# !( Granite Harbour # *# * UW / United Kingdom !( France Other AWS Cp Bird $+ !( Germany AGO "/ Commercial 0 100 km Evans Piedmont Gl $1 SPAWAR !( Italy !( Arelis !( !. Other US !( Japan

Coastline: ADD v4.1, 2003; Cartography: April 2016 Sam Batzli, SSEC, University of Wisconsin-Madison; Funding: National Science Foundation ANT-0944018 6"%7'+#%$8(*)(S%#3*"#%"5(,-.%#*"

V2(%04#>.*&I2#(%-) #11/,/$#>.*&),#M

6'), IU HIJ

!.01$%(=)b)JUYY *2(`

I-$-Q(/+$2'+" -".(*$2'+#5(BFFD

) !"#$"#%

3*-%$%-)X&>#01>.1)(/0+#1%),#(()2#$#&1%

6'), IU HIJ

N/&=)NX!3VU`JBXa[ L%0.*-=)J^_^IUYJJ N%(*$/>.*&=)U_)6, !.01$%(=)b) JUYY)*2(`

I-$-Q(?'"-'+$#(-".(*$2'+#5(BFAB

!*,M#0%):35)>*)GNX!O),#(()#&*,#$.%(

W.$>%0):35)+.%$-)+0*,)0%'.*&#$)1$.,#>%) ,*-%$)NX!3V)>*)(#,%)0%(*$/>.*&) #()GNX!O

,-+$%"(X*+Y-$2

* !"#$"#%

!*,M#0%),*-%$$%-) (&*8)-%M>")1"#&'%)>*)#$>.,%>0H

!*&4%0>)(&*8+#$$)#&*,#$.%()>*)+.0&)>".16&%(() 1"#&'%C)/(.&')#)+.0&) ,*-%$

,-+$%"(X*+Y-$2

!*,M#0%),*-%$$%-) (&*8)-%M>")1"#&'%)>*)#$>.,%>0H

I+*""%"< ,-&.(?-".

/0&".#'"(>'- '0E-80'"$

!*&4%0>)(&*8+#$$)#&*,#$.%()>*)+.0&)>".16&%(() 1"#&'%C)/(.&')#)+.0&) ,*-%$

,-+$%"(X*+Y-$2

! !"#$"#%

A0*&&.&')3#/-)P#&-C)O#(>)X&>#01>.1#

#; :35)+0*,)0%'.*&#$)1$.,#>%),*-%$ 2; !/,/$#>.4%):35)#&*,#$H)4(`)GNX!O

?'"-'+$#(-".(*$2'+#5(BFAZ

!*,M#0%),*-%$$%-) (&*8)-%M>")1"#&'%)>*)#$>.,%>0H

I+*""%"< ,-&.(?-".

/0&".#'"(>'- '0E-80'"$

!*&4%0>)(&*8+#$$)#&*,#$.%()>*)+.0&)>".16&%(() 1"#&'%C)/(.&')#)+.0&) ,*-%$

,-+$%"(X*+Y-$2

#+ !"#$"#%

X,/&-(%&) :%#)%,2#H,%&>C)R%(>)X&>#01>.1#

#; :35)#&-)A 2; 35)?):35)@ A 1; !/,/$#>.4%)35

>&$$'+G'8 -".(*$2'+#5(BFAJ

["%)\"#0-%0]) 1#(%=) G0%%&$#&-

V/>/O,WI!>

## !"#$"#%

JZ)H%#0()*+).1%)("%%>),#(()1"#&'%()+0*,)GNX!O

!"#$% &'()*+,,%-./01' !"#$%&'()*+,,%+'-%.2'(**%-./01'

/"$-+3$%3(%3'(#2''$ M+''"G-".(%3'(#2''$

T'G%3*<"- -".(*$2'+#5(BFAU

:35).()1"#$$%&'.&'= &*>)*&%C)2/>)>"0%%)2#$#&1%(D

E1%)("%%>),#(()2#$#&1%)935; 35)?):/0+#1%),#(()2#$#&1%)@ A.(1"#0'% FG> H0IJK

:/0+#1%),#(()2#$#&1%)9:35; :35)?)L0%1.M.>#>.*&) @ :/2$.,#>.*&) @ N/&*++)I O0*(.*& FG> H0IJK

P.Q/.-) 8#>%0)2#$#&1%)9PR5; N/&*++)?)N#.&)S)!*&-%&(#>.*&) S)3%$>)@ N%+0%%T.&')@ N%>%&>.*& FG> H0IJK

:/0+#1%)%&%0'H)2#$#&1%)9:O5; IU 3)?):R&%> S)PR &%> S)7)S)P)S)G( FR), K

KL(H-&G(M'$$8(,&#'&0

#& !"#$"#%

G0%%&$#&-)#/>*,#>.1) #&-),#&&%-)8%#>"%0) (>#>.*&(

/.-1$'.()+*0(MN6>

cI>0#&(%1>C 8%(>)G0%%&$#&-

S10

V/>/O,WI!>

#' !"#$"#%

>UQ(UFF(0(-#G G0%%&$#&-)(/0+#1%) %&%0'H)2#$#&1%

>DQ(AFFF(0(-#G

>\Q(AUFF(0(-#G

>$-$%*"(>D(9AFFF(0(-#G: >AFQ(A]UF(0(-#G

=*&+$'#8Q(H'$'+(@&%1'+# ,&""'['

N%'.*&#$) 1$.,#>%) ,*-%$

JJ)6,)0%(*$/>.*&

Impact of summer (a) (b) snowfall events on GrIS SMB

(c) (d) Title Page 3*-%$)%4#$/#>.*&=) (/0+#1%)%&%0'H) TCD Abstract Introduction 2#$#&1% 9, 1177–1208, 2015Conclusions References

Tables Figures Impact of summer UFF(0(-#G AFFF(0(-#G (a) (e) (b) (f) snowfall events on J I GrIS SMB J I B. Noël et al. Back Close

Full Screen / Esc

AUFF(0(-#G(c) A]UF(0(-#G(g) (d) (h) Title Page V*^G(-".(*$2'+#5(BFAU Printer-friendly Version Figure 5. Absolute values of observed and modelled turbulent and net shortwave/longwaveAbstract Introduction 2 ﬂuxes (Wm ) at station (a) S5 for 2004–2012, (c) S9 for 2004–2008, (e) S9 for 2009–2012Conclusions and ReferencesInteractive Discussion (g) S10 for 2010–2012; di◆erence in modelled and observed surface albedo and surface melt 2 Tables Figures energy (Wm )3*-%$%-)%4#$/#>.*&=) at stations (b) S5, (d) S9, (f) S9 and (h) S10 for the same periods, respectively. (/0+#1%),#(()2#$#&1% (e) (f) 1205 J I

6'),IU HIJ J I

Back Close

Full Screen / Esc

(g) (h)

Printer-friendly Version Figure 5. Absolute values of observed and modelled turbulent and net shortwave/longwave 2 ﬂuxes (Wm ) at station (a) S5N/&=) forNX!3VU`ZB 2004–2012,GN (c) S9 for 2004–2008, (e) S9 for 2009–2012 and Interactive Discussion L%0.*-=)J^deIUYJf (g) S10 for 2010–2012; di◆erenceN%(*$/>.*&=)JJ) in6, modelled and observed surface albedo and surface melt 2 energy (Wm ) at stations (b)V*^G(-".(*$2'+#S5, (d)5(BFADS9, (f) S9 and (h) S10 for the same periods, respectively.

1205

#$ !"#$"#%

G0%%&$#&-)1/,/$#>.4%),#(()2#$#&1%)9J^deIUYJf;

T-"(.'"(R+*'['(-".(*$2'+#5(BFAD

3*-%$$%- G0%%&$#&-).1%)("%%>),%$>B0/&*++) 9J^deIUYJf;

!"#$%&'())(*+,(- 7& *:/4;-&"<&-=>7-?<$-34-56 7& &!!! @:.6::CDAE;-?7-&<)-34-56 7& 567,0/#1&'0"*$/,'$,"2 @0AB..;-'<$-=>7-"<#-34-56

%!! 8 7& !"#$ $!!

%&'())

*+,,-./01-234-56 #!!

%")*""+,'- "!!

.&*)/0"12&3#,4/$,(' ! &'$! &'$( &')! &')( &'%! &'%( &''! &''( "!!! "!!( "!&! "!&( 9:+6

T-"(.'"(R+*'['(-".(*$2'+#5(BFAD

#% 9/15/16

Modelled Greenland ice sheet mass balance (1958-2014)

Trends 1991-2014 Discharge: 6.7 +/- 0.4 Gt yr-1 SMB: -10.7 +/- 2.9 Gt yr-1 Mass balance: -17.4 +/- 3.0 Gt yr-1 600 Solid ice discharge

400

) Surface mass balance -1

200

0 Mass flux (Gt yr

-200 Ice sheet mass balance

Typical uncertainties -400

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Year

Van den Broeke and others, 2016

Byrd glacier, Antarctica

17 !"#$"#%

E,M#1>)*+),*-%$)0%(*$/>.*&)*&)&%>)(&*8)#11/,/$#>.*&

Lenaerts and others: Impact of model resolution on simulated wind, drifting snow and SMB 823 Byrd glacier basin

Fig. 2. Modeled annual mean (2009) 10 m wind speed (contours) and direction (arrows) in RACMO/5.5 (left) and RACMO/27 (right). The locations of the four AWSs in Figure 4a–d are indicated by their respective letters. ?'"-'+$#(-".(*$2'+#5(BFAB

Drifting snow measurements are sampled at the Talos 5.5 km. In our simulation the strongest winds are found in the Dome TALDICE drilling site (72◦ S, 159◦ E, 2315 m a.s.l.; glacial valley of Reeves Glacier, with a maximum wind speed 1 http://www.taldice.org; see Fig. 1 for location). Measure- of 20.5 m s− , and, to a lesser extent, David Glacier near TM ments are obtained with FlowCapt driftometer sensors the Italian base (Mario Zuchelli) in Terra Nova Bay ( 75◦ S, ∼ produced by ISAW Outdoor Environmental Monitoring 163◦ E), and Byrd Glacier (81◦ S, 158◦ E). The occurrence (Chritin and others, 1999). The instrument is composed of these maxima is supported by results of Bromwich and of four sensors: two of these are placed 0.2 m above the others (1990), who showed that Reeves Glacier is the primary snow surface, and the other two 1 m above the surface. route for katabatic winds, and David Glacier is an important These instruments provide inaccurate estimates of the snow secondary outflow valley. transport fluxes (Cierco and others, 2007), but do provide a Lenaerts and others (2012b) showed that RACMO/27 realisticE,M#1>)*+),*-%$)0%(*$/>.*&)*&)&%>)(&*8)#11/,/$#>.*& estimate of the occurrence of drifting snow. underestimates high wind speeds in regions with complex Modeled SMB is compared with observations described topography. Figure 3 illustrates that in these regions, wind by Agosta and others (2012). The observations originate speeds in RACMO/5.5 agree better with observations. The 1 from a 150 km long stake line that runs from the coast of root-mean-square error (rmse) decreases from 5.7 m s− to ∼ 1 Terre Adelie´ to the southwest (0–1800 m a.s.l.). Comparing 3.6 m s− , and the mean bias between model and obser- 1 1 RACMOByrd to theseglacier observations basin can be regarded as a stringent vations drops from 4.3ms− to 1.4ms− . Nonetheless, − −1 test for model performance, because the stake line covers the the extreme wind speeds (>15 m s− ) in Terre Adelie´ remain strong SMB gradient between the relatively mild and windy underestimated. coastal climate and drier and calmer conditions inland at Figure 3 only shows long-term mean near-surface wind high spatial resolution (100 data points). speeds. Drifting snow processes, however, are usually connected to short-lived wind speed maxima. To evaluate the model results at higher temporal resolution, Figure 4 shows the daily mean 10 m wind speed from RACMO/5.5, RESULTS RACMO/27 and from available AWS observations. Due to Wind climate limited temporal coverage and large gaps in the data, these Figure 2 compares annual mean 10 m wind speed of AWSs are not included in Figure 3. At daily resolution, RACMO/27 with RACMO/5.5. Although RACMO/5.5 obvi- RACMO/5.5 shows clearly higher maximal wind speeds than ously shows much more detail, the regional patterns are RACMO/27 at all stations, except for Sitry (Fig. 3b), where 1 similar. We find four areas of strong (>14 m s− ) winds: topography is smooth and the model agrees very well with three over outlet glaciers (Byrd, Mulock and Reeves/David the observations, even at 27 km resolution. The other three Glaciers) in the Transantarctic Mountains, with several jets stations are known to be major confluence areas, where 1 above 10 m s− , and one in coastal Terre Adelie,´ with the katabatic wind accelerates due to the convex shape of 1 a maximum wind speed of 16 m s− at 69◦ S, 143◦ E, the glacier valley (Bromwich and others, 2000). At Larsen ∼ 1100 m a.s.l. On the smaller scale, the RACMO/5.5 wind Glacier, the modeled timing and frequency of wind speed field shows distinct features. Maximum wind speeds are maxima agree very well with the observations, whereas at higher and occur closer to the grounding line. Relatively Priestley and David Glaciers, observed wind speed maxima narrow (<20 km) glacial valleys, in which the katabatic wind remain largely underestimated, also by RACMO/5.5. The speeds converge and accelerate, are much better resolved at intense local katabatic flow at these locations is likely driven

?'"-'+$#(-".(*$2'+#5(BFAB

#* !"#$"#%

E,M#1>)*+),*-%$)0%(*$/>.*&)*&)&%>)(&*8)#11/,/$#>.*&

?'"-'+$#(-".(*$2'+#5(BFAB

O&"#&1%- -H&#,.1#$ -*8&(1#$.&'

N/&=)NX!3VU`ZBXL L%0.*-=)J^_^IUYJf N%(*$/>.*&=)d`d)6, !.01$%(=)b) JUY)*2(` =*&+$'#8Q(,'G32%*+(7-"(S'##'0

#! !"#$"#%

O#(>)G0%%&$#&-=).&-.4.-/#$) '$#1.%0()&*>)0%(*$4%-

JJ)6,)0%(*$/>.*& J)6,)0%(*$/>.*&

V*^G(-".(*$2'+#5(BFAD

O&"#&1%- -H&#,.1#$ -*8&(1#$.&'

:35

=*&+$'#8Q(S%GG'0(K-"(7-"(.'(R'+<

&+ !"#$"#%

O&"#&1%- (>#>.(>.1#$-*8&(1#$.&'

NX!3VU`Z)9JJ)6,;C)O#(>)G0%%&$#&- :>#>.(>.1#$$H)-*8&(1#$%-)>*)J)6, :35)9,,) H0IJ;

9R-#'.(*"(M!,H5(X*Y-$(-".(*$2'+#5(BFAJ:

V*^G(-".(*$2'+#5(BFAD

3*-%$%-)G0%%&$#&-) (/0+#1%),#(()2#$#&1%

6'),IU HIJ

N/&=)NX!3VU`ZBGN L%0.*-=)J^deIUYJf N%(*$/>.*&=)JJ)6, V*^G(-".(*$2'+#5(BFAD

&# 9/15/16

Outlook

Observations Focus on accurate radiation and turbulence measurements! New generation of autonomous weather stations New generation of satellites: GRACE-2, ICESat-2

Models Further improve existing atmospheric climate models (clouds) Further improve existing snow models (heterogeneous percolation) Move to global model systems (coupled ice sheet models) Improve prognostic albedo schemes (dust, black carbon, bio-albedo)