J ournaloJGlaciology, Vol. 44, No.

147, 1998

A 1 year record of global radiation and albedo in the ablation

zone of Morteratschgletscher, Switzerland

J. OERLEMANS, VV. H . KNAP

InstituleJor N!m'ine and AtmosjJheric R esearch, Utrecht Universil)" ,'VL-3508 TA Utrecht, T he Netherlands

ABSTR ACT . vVe analyse data o n sola r radiation m eas ured with an a utomatic weather
stati on on M o rteratschgletscher, Sw itzerland, for th e p eri od I O ctob e r 1995- 30 Septem-
b er 1996. Th e stati on is in the lower a bl ation zone. Du e to shading by surrounding moun-
ta ins and atmospheric attenuation, on ly 49 % of th e annual extraterres trial irrad ian ce
(mea n: 292 VV m 2) riac hes the g lacier surface. About 48% o[ th is is a bsorbed at the sur-
[ace (mea n: 79 W m ; a nnu al a lbedo 01' 0.53).
\Vc prese nt a simpl e albedo sc hem e for use in glacier mass-ba lance models. We fit th e
m odel to th e I yea r datase t by o ptimi z ing five co ntro l para mete r (o ptim al va lues in
brac kets): a lbed o of snow (0.75), a l b ed o of firn (0.53), a lbedo of ice (0.34), e-folding constant
[or effect of agell1g on snow albedo (21.9 days ) and e-fo lding co nsta nt fo r effect o[ snow
d epth on a lbedo (3.2 cm ). The input consists of da ily a lb edo, snow de pth a nd date of snow-
fa ll events. Th e co rrel ation coeffi cie nt between ob erved a nd simul ated a lbedo is 0.931, th e
corresponding rm s difference b eing 0.067.

1. INTRODUCTION
To und ersta nd how ablation o n a g lacier may ch a nge if
elim ate ch a nges, onc needs to know the components of th e
surface-energy b a la nce. It is ge ne ra ll y accepted th a t, a t least
on mid-l atitude \'a ll ey glaciers, the most importa nt pro-
cesses deli ve r ing m elt energy in summer a rc absorption of
sola r radi ati on a nd turbu lent exc h a nge of sensible h eaL. On
most glaciers, sola r radiati on typi ca ll y prov ides 75 % o f th e
mclt energy, a ltho ug h on th e lower pa rts of m a ritim e
g lac iers thi s m ay b e closer to 50%).
In recent years, g lacier mass-balance models h ave been
de\'eloped tha t a re based on a calc ul ation of a ll energy
transfers between atm osphere and glac ier surface (e.g.
Greuell and O eriem a ns, 1986; O erl em ans, 1993; O er lem ans
a nd Fortuin , 1992). R adiati\'e a nd turbu lent energy Ouxes
a re calcu lated from clim atologica l data by schem es used
widely in bo unda r y-l aye r meteoro logy. In th e course of
timc, th e usefu lness o[ such schem es has been tes ted with
fi eld ex periments o n glaciers in w h ich th e compone nts of
the energy bala nce were meas ured (e.g. Ambac h, 1979;
K uhn, 1981; Munro, 1989). Such stud ies were usua ll y p er-
form ed O\'er a short period during th e melt seaso n a nd at a
single locati o n. I n more rece nt yea rs, detail ed info rm a ti on
about altitud ina l g r adients has bee n obtained as a res u lt of
ex perim ents carried o ut simu lta neo usly at a number o[me-
teo rologica l sta tion s (e.g. O erlem a ns a nd Vug ts, 1993; Va n
den Broeke and others, 1994; Greue ll a nd others, 1995). H ow-
ever, these experim ents we re a lso restricted to the summ er.
In spite of the e nergy-bala nce work done so fa r a nd th e
sig nifica nt increase in our knowl edge thi s has broug ht
about, there is sti ll a need for meteorologica l d a ta on
glac iers coll ec ted over longe r p er iod s. In partic u la r, th e
parameter ization used for albed o in mass-bala nce m odels
needs furth er tes ting with data th a t cover both th e tran s-
it io n from snow to firn a nd to glacier ice, as well as the tra ns-
ition fr om a barc-ice s urface to snow cove r in th e [all.
It is, however, no t easy to obtain such data. Abla tio n
zo nes a re d iffi cult to wo rk in througho ut th e yea r, because
of melt water, high melt rates ca using co ntinu ous changes in
the surface reli ef a nd d iffi cult access in earl y winter (da nger
of avala nches and crevasses). Furth ermore, [or budgeta r y
a nd logistic reasons, it is not feasible to have a permanentl y
ma nn ed station on a g lac ier tongue.
Neve rtheless, th ere a re a few places where ome of th ese
difficu lties can be ove rco me; onc such p lace is ~ Io r te r a tsc h ­
glc tsch er, Switzerla nd (sce Fig. I). Thi s g lac ier flows in a
north e rl y di rection, is a b o ut 7 km long a nd has an area of
16.4 km 2. Altitude ra nges from 4049 (P iz Bernina ) to a b o ut
2100 m (glacier fro nt ). ccess lO th e tongue is poss ible in
a lmost a ll conditions. A m eteorologica l sta tion was in sta ll ed
here in September 1995 and has bee n fun ct ioning ever si nee.
I n thi s paper, we present a n a nalys is of the solar ra d i-
ation m easurements for th e peri od I O ctob er 1995- 30 Sep-
tember 1996. Wc fi rst present a bri ef d escript ion of the
instrumental set-up a nd then disc uss ce rta in features o f the
data. Next, a simple m od el for calc ul a ti ng the dail y a lb edo
is introduced. As input, it uses snow d epth and dates of
snow fa ll events. Th e m od el is filled to th e a lbedo d ata b y
opt il11.i z ing fi ve co ntrol p a rameters, na m ely, a lbedo ofrr es h
snow, a lbedo of firn, a lbedo of ice, e-fo ld ing consta nt for
effec t of ageing 0 11 snow a lbedo, e-foldin g co nsla nt for effect
of sllow depth on a lb ed o.
2. THE INSTRUMENTS
Th e a uto matic weather station (AWS ) is loeated on th e ton-
g ue of M orteratschg le tsch er, abo ut 500 m fr om the glac ie r
snout at a n elevati on of2 104m (as obtain ed from GPS meas-

231
Downloaded from https://www.cambridge.org/core. 21 Aug 2020 at 14:09:03, subject to the Cambridge Core terms of use.
 J ournal rifC/aciology
Fig. 1. A view over A1ortemtsclzgletsclzel; Switzedand. The location rif the automatic weather station is indicated by an arrow.
( Photogmph ~Y rv Haebedi, September 1989)
urements on 29 September 1995). It is placed o n a n homoge-
neo us pa rt of the glac ier which slopes graduall y (about 5°).
Th e AWS sta nds freely on the ice and sinks with the
melting surface. The di sta nce between senso rs and surface,
being 3.5 m , thus remain s approximately co nstant when
snow cover is absenl. In winte r, the height of the sensors is
3.5 m minus th e snow depth a t the AWS site. For the year
considered here, th e max imum snow depth was on ly
62 cm, whi ch is onl y half of the long-term average (accord-
ing to inform a tion from local mountain g uides ). Conse-
quentl y, the height of the se nso rs varied b etween 2.9 and
3.5 m.
The AvVS is equipped with se nsors for temperat ure (as-
pirated ), air press ure, wind sp eed a nd wind direction, a nd
with upwa rd a nd downwa rd looking pyranome ters. In ad-
dition, snow temperatures are meas ured. D ata are sampled
eve ry 2 min a nd then co nve rted into h a lf-hourly mea n
values a nd stored on a Campbell CRIO da ta logger. Power
is uppli ed by a sol ar panel combined with lithium batteri es.
The relative surface height is obtained from a n acoustic sen-
sor attac hed to a separate con struction drilled into the ice.
This all ows m on itoring of m elt and snow acc umul ation.
Stake readings were also carri ed out during r eg ular visits
to th e sta tion .
Th e Aa nderaa pyranometer is not th e most acc urate
instrument on the market but it is nevertheless suitable in
view of errors inherent in m easuring using an Th e A" rs.
potenti ally important erro r sources arc snow, co ndensati on
a nd riming o n the instruments, a nd tilting o f th e mas t. The
field of view of the downward-facing pyra nom eter is rel-
ativel y la rge, because the instrumenr is be tween 3 a nd
3.5 m above the surface.
It is poss ible to determin e global radiat ion b y using the
upward-looking senso r by partitioning the incom ing racli-
ation into direct a nd diffuse components, and ass uming th at
232
Downloaded from https://www.cambridge.org/core. 21 Aug 2020 at 14:09:03, subject to the Cambridge Core terms of use.
the two-stream a pproxim ation is valid . Howeve r, the parti-
tioning is the problem, since no obsen'at ions of cloud cO\'er
a nd reflectivity o f surround ing slopes are a\·ai lablc. "Vc have
chosen to correct the d ata direc tl y by comparing th cm with
th e data from ca refull y calibrated Kipp and Zonen p yran-
ometers opera ted for 4 weeks in spring ve ry close to th e
AWS site. The procedure is desc ribed in the Append ix.
The station has been operational since 29 September
1995 and was vis ited a nd observed o n 16- 17 December (im-
m edi ately after a 30 cm snowfa ll ), 19 February (duri ng light
snow), 19 M a rch, 18- 19 Apri l, 12- 13 M ay, 16 June- I Jul y, 31
Jul y- I August, 6- 12 September. So, in spite of t he fact that
the station was not m anned, m a n y obsen 'ations on its per-
formance a nd on the surface conditions are avail a ble.
During th ese visits, riming or condensati on was ha rdly
eve r observed. A ccording to mountain guides working in
the region, fog gy conditi ons on th e lower part of th e g lac ier
do not occur frequentl y, so possibl y the error res ultin g from
co ndensati on is no t too large, especially if dail y m ea ns are
considered.
The tilt of the m ast was measured on all the visits men-
tion ed above. In Scptember 1995, the station was placed
horizonta lly. On 16 December, th e tilt appeared to be 6° to
the north . This must have been due to the considerabl e
a m ount of m elt (about 0.4m ) in O ctober 1995. During the
winter, spring and summer the tilt did not change more than
a few degrees. In O ctober 1996, th e tilt was again m easured
acc urately, still g iv ing a value of 6 ° in a northerly direction.
With regard to the a lbedo m eas urements, it is fortun ate
th at the shading a ngle of th e surrounding mountain s, typi-
cally 30°, is sign ificantl y la rger than the tilt of the m as t.
Consequently, direct radiati on can never enter the down-
wa rd-looking p yranometer. This means that the potentiall y
largest error so urce is absent. In thi s paper, we d efin e the
a lbedo with resp ect to a hori zonta l surface. Therefore, all
 Oerlemans and KnajJ.' Global radiation and albedo at ,I IOl'teratschgletsc/zer

caleula tio ns have been done using d a ta reduced to the ho ri-
zol1la l pl a ne as described in the Append ix.
3. GLOBAL RADIATION
Fo r a proper inte rpre ta tion of the m eas uremeJ1lS prese m ed
here, some basic co nce pts on sola r ra di a tion need to be cl a r-
ifi ed. Th e sola r co nstant S is defin ed as the energy flu x of
ra di a ti o n ori gina ting from the Sun, p e rpe ndicul a r to th e so-
la r bea m, fo r th e a nnu al mean E a rth Sun di sta nce (e.g.
Li o u, 1992). Th e c urre ntl y acce pted va lu e is 1368 ± 3 VV m 2
For r eference, it is useful to defin e the ex tra-tenT tri a l irra-
di a nce Qex as
Qcx = SsinA .
H ere, A is solar eleva tion. Qcx thus re prese nts the flu x of
e n ergy through a surface of unit a rea p e rpendicul a r to th e
E a rth surface a t th e "top" of the a tm os phere. An cfketive
tra nsmi ss i\'ity Tcrr ca n now be introduced by rcla ting th e
g loba l ra di ati on C to Qex :
C = TcrrQcx .
It should be stressed th at here g lo ba l radi ation refe rs to
the flux impinging o n an im agina r y hori zoJ1la l surface
thro ugh a point o n th e sloping glacier surface.
Th e result of a number of processes is absorbed in Terf' .
Fi rst of a ll , a tm osphe ric processes, th a t is, scatterin g a nd a b-
sorptio n of sola r ra di a tion by a i r, ae roso l a nd cloud s, m a ke
Tl'il sig nifica ntly sm a ll e r tha n un ity. R ef1 ection fr om the sur-
ro unding slopes, o n the other ha nd , no rm all y lea d s to a n
inc rease in globa l ra di a ti on, i.e. a n inc rease in Terr . In prin-
c iple, Tell can th erefo re be la rge r th a n unit y but thi s ra re ly
ha ppe ns.
Gl oba l radi a ti o n ca n be pa rtitio ned into direct ra di a ti o n
(sola r bea m imping ing on the surface ) a nd diffuse ra diati o n
(coming from a ll o th e r d irecti o ns). Th e proporti o n of dif-
fu se ra di a tio n inc reases with clo udiness a nd reOectivit y of
th e surro unding slo pes. A full treatm e nt of wha t ha ppens
with sola r ra d ia ti o n in a glacieri zed \'a ll ey of given g eom-
e tr y therefore beco m es a co mpl ica ted m a tter, so co mpl i-
ca ted th a t o nl y co mputer time-co ns uming ~\lo nt e C a rl o
m e th od s can deal with a ll as pec ts. Thi s is not feas ihl e fo r
the stud y of the ene rgy ba la nce of a g lac ie r surface, in whi ch
o th er processes such as turbul ent flux es introduce sig nifi-
ca nt unce rt a iJ1lies.
A fr equentl y m a d e ass umpti o n in r a d ia ti\'e-tra nsfer c a l-
c ula ti o ns is th at th e diffusi\'e sola r ra di a ti o n, including ra d-
ia ti o n reOected by th e glac ier surface, is isotropic. Thi s
impli es that th e a mo um of reOected r a dia ti o n meas ured
with a d o wn-facing py ,'anom e ter should no t depend on th e
tilt of th e instrument, unless it is tilted so much that direct
radiati o n ca n ellter th e do me when th e Sun is \'er y low
(l\Ia nnstein, 1985).
\ Vc no w turn to the m eas urements. A s a n exa mple, Fig-
ure 2 sho ws the record of g lobal a nd refl ected radiati on fo r
the first 8 d ays of April 1996. A few ce ntim e tres of snow fell
on 2 April. Th e ex tra-te rrestri al irradi a nee fo r I day is a lso
shown fo r reference. Th e la rge cut-off in g loba l radi a ti o n ,
beca use of shading by th e surround ing m o unta in s, is clea rl y
see n. This, of co urse, is t y pica l for a no rth-fac ing glacier in
the No rthern H emi spher e. 8 April appa re ntl y had va ri ahle
lig ht cl o ud s a nd at som e in sta nt a peak g lo ba l radi ation ve r y
(1)
close to the extra-t errestri a l irradi a nce. Thi s must h3\'e bee n
due to re fl ec ti ons be twee n clo ud, glacier sur face a nd sur-
roundin g slopes.
D a ily m ea n va lues of C for th e enti re yea r of measure-
ments a re plotted in Fig ure 3, together with t he extra-terres-
tria l irra di a nce. Air temperatures (a lso d a il y mea ns) a re
(2) shown fo r reference. The m ost striking feature in the ra di-
ation record is the la rge day-to-d ay va ri a bility in th e
summe r h a lf~ yea r compa red to the w inte r ha lf-yea r. \ Ve
bcli eve tha t thi s is assoc ia ted with th e prese nce of snow o n
the \'a lley slopes. The effec t of mu ltiple refl ec ti o ns on globa l
radi a tio n is fa r more impo rta nt when th e slo pes a rc co\'e red
with sno w tha n when they a re not. Th erefo re, in cases where
the slo p es a rc covered with snow, mu ltiple reOecti o ns
decrease the contras t be tween sunn y a nd cl o ud y days. It is
possible th a t a nother process, associa ted with differe nt
clo ud c h a rac teri stics, c ontributed to th e diffe re nce betwee n
summ e r a nd winter. In w inte r, cloud s a rc p,'oba bly opticall y
thinne r.
A s no ted ea rli er, inte rceptio n of sola r ra di a ti on by sur-
roundin g m ounta ins stro ngly reduces th e incoming ra di-
ati o n, no ta bly from eas te rl y (ea rl y m ornin g ) a nd weste rl y
(late a ft e rnoo n) direc ti o ns. C onseque ntl y, the re is a la rge
differe nce be twce n da il y m ean globa l ra di a ti o n a nd ex tra-
terres tri a l irradia nce.
Ta ble I lists monthl y m ea n \'a lues of C , Qt'x a nd Tofr (th e
las t three columns a re di sc us ed later). In summer, abo ut
ha lf o f rh e p o tenti a l sola r e ne rgy reaches the g lacier surface.
In winte r, th e fr ac ti on is eve n lower.
4. ALBEDO
Fig ure 4 shows a ll the m eas urements of inco ming a nd rc-

1200 , -_ _-1 - - - global

')' ......... reflected .,\~ ~xtra-terrestrial irradiance
E 1000 . - . - extraterr. ..................... " ..... -. ""''''''T \ ....................................................................................... ;.................. j
~ '--,-------,--'
....
c: 800 ~.

.Q
Cil 600
:0
~
400
~
0
en 200

0
2 3 4 5 6 7 8
April 1996

Fig. 2. Example qf measured global mdiation ( solid) and l'if/ecled solar radiation ( dashed). Shown are hatJ-ILOlI rly mean vallles.
Extra-terrestrial irradiance is slzownJol' reference.

Downloaded from https://www.cambridge.org/core. 21 Aug 2020 at 14:09:03, subject to the Cambridge Core terms of use.
233
 Journal qfClaciology

15
. : . :
10 ·······························t··································t··································r··································1························· ....

j::j ~t:: ~: -:!r

[)
e...-

~
~
:::J

Q)
0 :
...........................; .................................. .; ......
Co -5
E
~
... -10 ............................. :................................. ·"1".... ············.. ················1····················.....................
«
-15 . . . . . . . ................ Toec. i"..................·_..........F·eb·i ........ .. ............·Ap·r..i.... "j..........................·j'uii\ . . . .................. . . ....·L.................................L ...

-20
500

. .
')'E 400 ...................... ·...... ··• ........................· ...... ·l... . extra-terrestrial
irradianfe ~ """"""'r ..................... · ········i· · ···· ·· · · ·························~·········· ·····················1······

300 ............................... ,; ..................................................................... ,.............. ·................ ·i .................. · ·..

!
~
"
200 ........ .. ................. '1' .......... ......... . . . . . ........................ . . .
I
ao 100 ........... ·l··· .... ·· ....................... ................ ..
~

o .............................. + Dec 1.............................. Feb 1............ +. . . . . . . . . APr 1..... j...........................Jun ;1............................... + 1""""'''''''''''''''''Fe p 30

o 50 100 150 200 250 300 350

Day number (0 = 1 October 1995)
Fig. 3. Daily mean values qfair temperature (upper panel, smoothed curve also shown) and global radiation (lower panel).

Dected r a diation . Each dot represcnts a ha lf-hourly mean ing to the upward-l ooking senso r, whereas the downward-
value (total number of d ata p oints: 17 568). There is a clear loo king sensor remains free of snow. The A\vS was \'isited
grouping in the diagram. Apparently, cha rac teristic albedos once in light snow but then th e sensors were clean. During
for snow/Ern and ice a re em erging. a visit after a heavy snowfall (35 cm , 18 October 1996), the
O ccasionally, albedo values larger than one occur. effect co uld be clearly observed (upward-looking sensor
Closer inspection of the d a ta in relation to the snow-depth covered by a few centimetres of snow, downward-looking
record (di sc ussed in more d eta il later) shows th at this hap- sensor free of snow ). In any case, such measuring err ors havc
pens during snowfall a nd must be due to snowfl akes ad her- only a small effect on the over a ll pattern.
On most days, a lbedo shows some variation during the
Table /. Solar-radiation characteristics in monthly mean day. "Ve hesitaLe to give a full a na lysis of thi s, because severa l
values. G is global radiation, Qcx is extra-terrestrial irradi- [actors that may cause this variaLion (varying tilt of sensors,
ance, Teff is transmissivit)', A is absorbed solar radiation, f is riming, snowdrift, etc.) co uld not be observed because the
absorbed solar radiation scaled by extra -terrestrial irradiance.
In this table, albedo is the ratio of monthly mean riflected to
global radiation 1200

'"E 1000
Month G Qcx r eff A Atbedo f
2 2
~
\V m 2 \Vm Wm c 800
0
.~

'5
October 1995 106 210 0.51 71 0.34 0.34 ~ 600
November 1995 57 138 0.41 26 0.54 0. 19 '0
Q)
Dece mber 1995 37 108 0.35 10 0.72 0.10 U
Q)
400
J anuary 1996 52 128 0.41 13 0.76 0.10 ~
Febru a ry 1996 98 191 0.51 29 0.70 0. 15 er:
I\[arch 1996 167 54 0.68 0.19
200
282 0.59
Apri11996 221 377 0.59 86 0.61 0.23
t-1ay 1996 236 4+7 0.53 11 6 0.51 0.26 0
.June 1996 282 476 0.59 194 0.31 0.41
.Jul y 1996 241 455 0.53 167 0.31 0.37 o 200 400 600 800 1000 1200
August 1996 174 390 0.+5 113 0.35 0.29 Global radiation (W m-2 )
September 1996 134 301 OM 72 0.47 0.24
Fig. 4. ALL half-hourly means qfglobal and riflected radiation
Year 151 292 0.49 79 0.53 0.24 in a scatter plot. Two clusters eme7ge, associated with snow
and ice.
234
Downloaded from https://www.cambridge.org/core. 21 Aug 2020 at 14:09:03, subject to the Cambridge Core terms of use.
 OerLemans and Knap: Global radiation and albedo at MorteratsclzgletsclzeT

station is unm a nned. From now on, the discussion will be iation A a nd th e effici en cy of abso rbing sola r radiation (1)
restricted to daily albedos. First, daily albedo is defin ed as defi ned as
the ratio of daily amount of reflected solar radiation to the
daily global r adiation. Note that this differs from th e d a il y (3)
m ean albcdo, which is the mean of a ll instantaneously d eter-
So, f indicates how much of the insolation at the top of th e
mined albedo va lues during daylight. Howevcr, wh en the
atmosphere is ac tu a lly a bsorbed at th e g lacier surface. In
Sun is ve ry low, this qu antit y m ay be influcnced greatly by
the a nnual mean, only one-quarter of th e extra-terres trial
m eas uring er rors. In view of this, dail y mea n albedo is not a
irradia nce wou ld be a bsorbed by a hori zontal surface on
very useful qu a ntity in analysing m eas urements and instead
the g lacier at the AWS site! The amount will be even less
the daily albedos should be st udied .
for the actu al glacier s urface, which h as at the AVVS site a
Figure 5 shows dail y albedos for the cntire yea r. For rea-
slope of abo ut 6° in a north erl y direction.
sons explain ed a bove, I day app ea red to have an albedo ex-
ceeding 0.9. For this day, its value was se t at 0.9. Also shown
in Figure 5 is snow d epth deri ved from the acoustic se nso r. 5. MODELLING ALBEDO
As expected , there is a clear re lation between a lbedo and
snowfall events (seen in the record of the acoustic sensor as In mass-balance m odels based on a calcu lation of the
a n almost stepwise increase in snow depth). energy balance, a lbedo should be a prognostic varia bl e
In ge nera l, snow albedo is lower in spring than in winter, determined by the n a ture and hi stOl-y of the melt process.
except when la rge amounts of snow fa ll (as on 21 - 22 June, In ea rli er studies, simpl e formulations for albedo have been
when a few centimetres of snow bring the albedo up to over used in which one or more of the following pa rameters plays
0.8). It a lso appears that ice albedo increases slowly in a role: snow depth, acc umulated melt, age of snow, snow
August a nd September. There a re three snowfall eve nts in density, distance to equilibrium line, ete. Because of th e lack
late summer (27 August, 12 a nd 24 September). In a ll cases, of d ata, such formul a tions have neve r been checked th or-
snow melted away within a few d ays but the a lbedo re- oughly. Although the present dataset is far from ideal and
mained rela tively high. Consequently, the monthly a lbedo co ncerns onl y a sing le p o int on a glac ie r, it provides a n op-
for September 1996 was sig nifi cantl y higher than for portunity to test som e of th e ideas b ehind th e alb ed o
O ctober 1995. Howeve r, O ctob er 1995 was exceptionally schel11.es.
sunny a nd warm (see also th e temperature cur ve in Figure H ere, we wi ll fit a n a lbedo scheme, described bri efl y, to
3). All the monthly albedos a re give n inTable l. the 366 da ily albedo values. Although one co uld arg uc that
The last two columns inTable I show abso rbed so la r rad- one loses informatio n by not looking a t h a lf-hourl y values,

0.8 ......................... .

o
0.6 .............................. ·!·····································t· ............................... ~ ..
"0
Q)
.0
.
Cii .-.~ ............... -_ ............. ····r ......................... + ......................................!.................... .

0.2
: Oec 1 Feb 1 Apr1 Jun 1 ~ug 1

0.8

0.6
E
.c:
15.. 0.4
Q)
"0
3: 0.2
0
c
CIl

0 50 100 150 200 250 300 350

day number (0 = 1 October 1995)

Fig. 5. Daily albedo ( ujJjJer jJanel) and snow depth ( lower panel), derivedJrom the acollstic sensor ( excelJt the jJeak marked • J
Jar which snow depth was arbitrarily described).
235
Downloaded from https://www.cambridge.org/core. 21 Aug 2020 at 14:09:03, subject to the Cambridge Core terms of use.
 Journal qfGlaciology

the additiona l informati on, nota bly on snowfall evelllS, w e re calcu lated to remO\'e noise. It was ass umed Lh at there
would not be acc urate enoug h for a stud y wiLh a high reso- was a snowfa ll on d ay (i ) if
lution in ti me. In the calc ula Lion, values o[ fiv e paramete rs
a re dete rmined in such a way that the rm s diffe rence (CT ) d(i) - d(i - 1) :2: 0.02 m. (6)
be Lween obsen 'ed a nd simu la ted daily a lbedos is mini-
H err, d( i) is th e dai ly mean snow depth. The value o[
mi zed.
First o[ a ll , we anticipa te that the albedo of th e snow- 0.02 m was chosen , because it is twice th e estim ated snow-
covered g lac ie r site at day (i ) w ill depend o n th e age of th e d e pth error for d a ily mea n \'aILles. Th e resulti ng snowfall
eve nts a rc shown in Fig ure 6. Ve r y probabl y, a few o f th e
snow a t the surface, so
snowfalls derived from the applicat ion o[ Equ atio n (6) to

OOSIlO)i) = OOfirn + (OOfrsllo", - OOfirn)exp ( S ~ i). (4)

the snow-depth d a ta a rc not rea l a nd a few may b e mi ss ing.
This is the price one has to pay when using data from an
autom atic station .
Note that this express ion introduces three parameters: a It is not d ifficu lt to ca lculate th e optimal values for the
characterist ic a lbedo [or rirn (Cl'lh'n ), [or fresh snow (OOfrsllow) cOI1lrol pa ra m eters. \ Ve round that t he mutual d ep e ndence
a nd a time-scale t * deLermining how fas t the snow a lbedo of optimal control p a ra meters is sm a ll a nd that a unique
approac hes th e firn albedo aft e r a snowfall. In this equ ation, so lution co uld b e obLained by m inim izing CT. T h e results
S is the number of the day on which the last snowfall are given in Table 2 (fu ll model). A lthoug h the corre lat io n
occurred. Because of the d aily resolution used h ere, the co effi cient is significa l1l (0.93), CT is still fairly large (0.067).
a lbedo of fr esh snow should be interpreted as th e c ha racter- J udging from Figure 6, in which observed a nd calcul a ted al-
iSLic albedo [or snow that is n o t more th a n I d ay old. So, it bedos a re co mpa r ed , there a re at least two reaso ns. Around
may (and d oes ) differ from the albedo of a snow surface days 135 a nd 160, the effect o[ ageing is ap parently overesti-
during o r immedi ately after a snowfall. m ated in the model. A poss ible reaso n is that in these p eri-
Next, wc require a smoo th transition to the c h a racte ris- ods lig ht snowfa ll s w ere not revea led by the a nalys is o f Lhe
tic ice albedo OOicc when snow d e pth (d) is sm a ll. This can be snow e\·ents. A seco nd facto r conce rn s the tendenc y of the
ac hi eved by writing [or th e fin a l a lbedo
Table 2. Summmy qf model results. For each model, the best

00
(i ) _
- OOsnow
(i) + OO'.CC _ O!SIIOW (I) exp (-d) d* (5)
valllesJol' the con trol jJarameters are listed, as wpll as thp I'ms
difference ( CT) between simulated and observed albedo and
corresjJonding conelation codJicient (r)
Here, d* is a ch aracteristic scale [or snow depth . \ Vhen snow
depth is d*, the snow cO\'Cr co ntributes li e to th e albedo, the
°fl":- l1 o\\ G rim a icI' d' t' er
underl ying surface (I - li e). If snow depth eq uals 3d', th e
f ill d
underlying surface sti ll contributes about 5% to th e albedo.
Therefore, there a re now five control para meters:
Full model 0.75 O.S3 0.34 3.2 21.9 0.931 0.067
OOfrsnow, OOfin" O!ice, t* a nd d*. Before optimal \'aILles for these
On c a lbedo for sno\\' 0.70 0.3+ 2.9 0.912 0.076
can be calcu laLCd , the snow fa ll e\'ents ha\'e to b e defin ed. cl* = 0 0.71 0.5+ 0.3+ 0 21.3 0.891 0.08+
First of a ll, d a ily mea n \'alues of the snow-d epth measure- Two albedo "aILles onl y 0.67 0.3+ 0 0.882 0.OS7
ments from the acoustic se nso r (sampli ng eve ry 3 hours)

!!2

~o
cCl)

0.8 .. .... ~.................... .................~ ...... .

0
0.6
"0
Q)
.0
~
0.4 - - albedo obs.
••• •• ••• • albedo sim.

0.2 .......:. .. . ................ ~............ . ... .. . .. ........•.................................. ; ................................... .....

~

Dec 1 Feb 1 Apr 1 Jun ' 1

0
0 50 100 150 200 250 300 350
day number (0 = 1 October 1995)

Fig. 6. UjJPer panel: snowJaLL events derivedJl'Om the acollstic snow -depth measurements. L ower panel: a comparison 0/observed
and simulated albedo.
236
Downloaded from https://www.cambridge.org/core. 21 Aug 2020 at 14:09:03, subject to the Cambridge Core terms of use.
 Oerlelllans and A-na/): Glohal radiation and albedo at 1110rteralschgletscher

0.1,---.---.----,---.----,---,---. I I

0.09 ...... ,........... -

\:) 0.08

0.07
\ \ .. ~
~\ /
/
'-.. r-- "--(
<X <X!rsn
ice
0.06 L -_ _L -_ _-'---_ _--L_ _--'-_ _ _ _L -_ _- ' - - - - - l
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 10 20 30 40 50 o 0.02 0.04 0.06 0.08 0.1
albedo t* (days) d* (m)

Fig. 7 rills difference between simuLated and observed aLbedo ( 0' ) for vm~ying control IJammeters. ControL parameters were varied
one ~y onf. The minimum in each curve corresponds to the bestJit with lheJuLl model given in TabLe 2.

obser ved ice albedo to increase from the b eginning of Jun e takes pl ace, the percentage is higher, na m ely a bout 32%
to the e nd of September. (because . h ading by th e surrounding m o unta ins is less
"la im'estigate th e sta bility of th e a lbed o model, the co n- efTective th e n ). The order-o f-m ag nitude is pro ba bly typical
trol paramcter. were \'a ri ed around th e optim a l values. Fi g- for la rge r north-racing glac ie rs in the Alps.
ure 7 sh ows the res ult in term s of the rm s difTerence between Albed o was found to va r y strongly in time, notably in
obse r ved a nd simulated a lbedo (0' ). Ice a nd noli' albed os
a ppea r to be better defin ed th an the firn a lbedo. The cur ve 1200 r---,---,-----,-----,-----y----.---.
fo r the age ing time-sca le t * is rather fIat : it would not m a ke
much difJerence irt* 'were to be increa sed o r decreased by a 1 000 ........... ~
few d ays. The snow-depth scalc d* is 111 0re sharpl y de te r-
min ed. The implicati o n is th at, for a sn ow depth of 10 C I11 , '" 800
'E
95 % of the a lbedo is d e termi ned by the snow cO\'er and o nl y
5% by the underl ying ice surface. Thi s is in good agree m e nt ~ 600 .... ~ ...... (
··········!················r···········
0..
with th e th eo reti cal res ults ofWi scombe a nd Wa rren (1980). 0..

Onc m ay well wo nd er wheth er simpl e r a lbedo mod els ~ 400

wo uld g ive a simul ati o n of compa rable qu a lity. [n Ta bl e 2 iI
th e res ults of a ca lcul a ti on using simpler m odels arc sum- 200
ma ri zed . First, we co nsider a model with out age ing eflcc t
of the snow. So, a sing le value for snow a lbedo is used. Th e o
best res ult is obta ined with a snow a lbedo o f 0.70, the res ult-
-200 o 200 400 600 800 1000 1200
ing \'a lue of 0' being 0.076. One may therefo re co nclude th a t
Rin AWS (W m-2)
th e simul a ti on witho ut a n ageing efTec t is significantl y,
a ltho ug h not d ramatically, \\'orse.
Another simplificati o n would be to omit the snow depth
800
(i. e. put d* to zero ). Thi s leads to a n optim a l simul ati on with
0' = 0.084, clea rl y demon strating tha t it m a kes se nse to ta ke
snow d epth into acco unt ex plicitl y. Fin a ll y, a run was do ne "!~ 600
ill which o nl y two albed o \'alues were used: o ne for snow a nd E
o ne fo r ice, no acco unt b eing taken o f efTects of sno\\' de pth ~
0.. 400
or age ing. Now, the bes t si mul ati on th a t can be ac hi eved h as 0..
~
0' = 0.087, which is 30 % la rger th a n the res ult for the full
~
mod el. ~ 200
\ Ve m ade se\'era l furth er attempts to impro\'C the simul-
ati o n by ta king into acc ount air temper a ture and so la r e le-
vati o n. Thi s was no t successful a nd th e results of these 0
additi o nal ca lcul ati o ns a re therefore no t shown .
0 200 400 600 800
2
Rre!1 AWS (W m- )

6. CONCLUDING REMARKS
Fig. 8. .r1 com/Ja rison qfdataJ1'01rl the Aandema kvranometers,
Altho ug h a I yea r reco rd does not co nstitute a clim ato logy, mOllnted 0 11 the rI r I :5, all d th e k ijJjJ p-)II'allomelers, IJLaced lJer-
we m ay co nclude th a t, o n a n annu a l b as is, th e tong ue o f Ject~)' /w1'i;::ol1tal. Each dot relJresents a Iw!f-holl r£y mfan
M ortera tschgletscher a bsorbs lcss tha n 25% o f the sola r ra d- Ni/llf. Th e tol) jJanel shows global radiatio n and the bottom
iati o n ava il abl e at the top of the a tm os phere. Fo r th e l)al1el shows riflected solar radiatiol/. Period: /8 April
c~ month peri odJun e- Se ptember, wh en m os t of th e melting 7530 It unld 15 May 0930 h I 'T 1996.
237
Downloaded from https://www.cambridge.org/core. 21 Aug 2020 at 14:09:03, subject to the Cambridge Core terms of use.
 JournalofGlaciology

late spring and fall when snow fa lls every now and the n o n
bare ice. An albedo scheme with Gye control parameters is
able to reproduce the observation reasonably well if snow-
fall events a nd snow depth are used as input. From compara -
tive tests with simpl er versions of this scheme, we conclude
that the a lbedo model formulated here is suitable for simu-
lating the a lbedo. Its deg ree of compl exity is in balance with
the type and quality of the measurements.
The a lbedo scheme proposed here ca n be incorporated
in glacier mass-balance models, if basic information is ayail-
able on snow depth and frequency of snowfall events. We
hope that it will be possible to use this scheme for other
points on a glacier by adjusti ng only the ice albedo. For
many g laciers, the variation in the characteristic ice albedo
can be estim ated from high-resolution satell ite images (e.g.
optica l ch ann els of Landsat TM) acquired on a suitable day
in the second half of the ablation season (sce e.g. Kn ap a nd
ot hers, in press ).
REFERENCES
Amba ch, W. 1979. Zum \\'a rm eha usha lt des g ronl a ndi sc hen In la ndeises:
\'ergleichende Studie im Akkumulations- und Ablati onsgebiel. Pola/"-
jorscizung, 49 (1), 44--54.
Greuel l, \<". a nd.J. Oe rlemans. 1986. Sensitivit y studi es wit h a mass ba la nce
m odel including temperature profil e ca lcu lations in side the glacier. Z
Glelscherkd. Gla::.ialgeol. , 22 (2), 101 In
Greuell , W., :d. R. \'an den Rrorkr. W. Kn ap, C. Reij mer. P Smeets a nd I.
Struijk. 1995. PASTEX: a giacio-meleorological eX/Jerimenl on IIIe Paslerze
( AI/slria). Utrecht, Utrec ht Uni\·ersit y. Institute fo r tvlarine and Atmo-
sph eric Resea rch; Am sterda m. Vrij e Universir eil. fa cult y of Ea rth
Sciences. (Field Re po rt.)
Kn a p. W. H ., B.\v' Brock, j. O erlemans and I. C. Willis. [n press. Compar-
melting zon e of the West Green la nd ice sheet in summer. /3ol/lIrial) '-La),er
.I/eleorol.. 71 (4), 393- 413.
\\'iscombe. \\'.j. and S. G. Warren. 198 0. i\ model for th e sp ec tra l albedo of
snow. I. Pure snow. J ,l/mos. Se;' , 37 (12),2712- 2733.
APPENDIX
The output from the Aanderaa pyranol11eters was corrected
by a comparison with Kipp pyranometers. The Kipp pyran-
ometers were placed perfectly horizontal and close (a few
metres away ) to the site of the AWS. After looking at the
data (18 April 1530 h until 15 May 0930 h UT), it appeared
that the correlation between h a lf-hourly values from the
Kipp a nd Aanderaa sensors was very high (sec Fig. 8). The
systematic e rror in the incoming rad iation can be explained
by the 6° tilt (towards the north ) of the Aanderaa senso r.
In \·iew o f other possible sources of error in the Aanderaa
data (temperature effects, snow cover, riming a nd less opti-
mal cos in e response ), it was decided to correct the Aanderaa
data by a straightforward application of lin ear regression
equations for globa l and refl ected radiation. The regression
equations were calc ulated so that the rIns d ifference
between the Aanderaa and Kipp and Zonen data was mini-
mized under the cond ition that the mean \'alues for both da-
tasets were identical. An exampl e of th e correction
procedure is shown in Figure 9.
- - Aanderaa
1200 .. . .. .... Aanderaa corrected
!
- - Kipp
~ 1000

t..
N
iso n ofL a nd satTr-. l-derived a nd g round-ba sed a lbedos of H a ut G lac in 'E
d'l\ rolla, Swit zerl and. /nl. J Remole Sensing. ~ 800
Kuhn , M . 1981. Die Reakti o n del' Schneegrenze auf Klim aschwa nkunge n.
Z Glelscherkd. Glazialgeoi. , 16 (2),1980,241 254. c
~ 600
f .. 10
Li o u, K. N. 1992. Radialion and cloud prOfesses in Ihe almos/JII fre. New York,
Oxford Lini\'e rsit y Press. '0
~ 400 t
r-. l annstein. H . 1985. The interpreta ti on of albedo m easurements on a sno \\'
covered slope. Ilrch. Meleoro!' Geophys. Bioklimalol., u B, 36,73 81.
Cij
.c
t\ I
\
o
r-. lunro, D. S. 1989. Surface rou g hness a nd bul k hea t tra nsfer on a glacier: a 200 "H

compari son with eddy co rrelation. J Glaciol.. 35 (121),343- 348.

O erl ema ns, j. 1993. A model for th e surface bal a nce of ice masses. Part I.
A lpine glaciers. Z Glelscherkrl. Gla<;ialgeol., 27-28, 1991- 1992,63 83.
or--' 1 \ j \ J
----t--" ~

O erl em a ns, J. andJ. P. F. FOI·Lu in . 1992. Sensit iv it y of glaciers and sma ll ice 27 28 29
caps to g reenhouse wa rming. Science, 258 (5079), 115 117. Date (April 1996)
O erl emans, j. a nd H. F. Vug ts. 1993. A meteorologica l ex periment in the
melting zone of th e Gree nl a nd ice shee t. Bull. Am. ,l/eleorol. Sac., 74 (3).
355- 365.
Fig. 9. An exam/J/e of corrected global radiation. Th e signal
Van den Broeke, r-.t. R. , P. G. Duynkerkc and E. A. C. Henncken. 1994. from the Aanderaa pyranometer has been corrected to the Kipp
Heat. momentum and m o isture budge ts of the kataba tic laye r O\'e r the py ranometer.

MS received 14 Februa1Y /997 and acce/Jted in revisedform 17 November 1997

238
Downloaded from https://www.cambridge.org/core. 21 Aug 2020 at 14:09:03, subject to the Cambridge Core terms of use.