Jökull - 01.07.2003, Qupperneq 17

A calibrated mass balance model for Vatnajökull
Table 4. Correlation coefficients between observed and modeled mean specific mass balance for the drainage
basins where the mass balance has been measured. – Fylgni mældrar og reiknaðrar afkomu á nokkrum vatna-
svæðum.
Region Area (km ) Winter Summer Annual
Tungnaárjökull 491 0.86 0.81 0.93
Köldukvíslarjökull 330 0.65 0.89 0.98
Dyngjujökull 1239 0.77 0.92 0.96
western Brúarjökull 956 0.68 0.82 0.95
eastern Brúarjökull 780 0.65 0.90 0.97
Eyjabakkajökull 121 0.42 0.76 0.86
 varies strongly over Vatnajökull (third column).
In regions with high precipitation in the south and east
 is, depending on hypsometry, up to 70% higher
than in the drier regions in the northwest. The same
relation has been observed for a set of climatologi-
cally very different glaciers (Oerlemans and Fortuin,
1992) and our results fit well in this picture:  is
almost -0.8 m w.e./K for glaciers with high precipita-
tion and about -0.50 m w.e./K for drier glaciers. Thus,
the precipitation gradients over Vatnajökull cause the
sensitivity to temperature changes to vary consider-
ably over the ice cap. Differences in precipitation also
result in differences in  , but this sensitivity varies
less over Vatnajökull.
Under the influence of increasing concentrations
of greenhouse gases in the atmosphere, an annual tem-
perature increase of about 0.3 K per decade is ex-
pected for Iceland in the next decades (Houghton et
al., 1996). For Vatnajökull an external temperature
increase of 1 K implies a decrease of  with 0.71
m w.e. When we add an increase in precipitation of
5.3% for a warming of 1 K (Huybrechts et al., 1991),
the decrease is smaller but still 0.56 m w.e.
DISCUSSION AND CONCLUSIONS
We have presented a mass balance model for Vatna-
jökull which has been calibrated with an exten-
sive data set. This made it possible to study differ-
ences between the free atmosphere and the katabatic
layer. The calibration of the parameterization for  demonstrated that over Vatnajökull,  is better de-
scribed as a function of  and e than as a func-
tion of  and e. This is a consequence of the mod-
erate size of Vatnajökull, which allows advection of
relatively warm air over the ice cap. This air strongly
determines  , as the katabatic layer is not very thick.
Nevertheless, the katabatic layer is well enough devel-
oped for  to deviate significantly from . The
model presented in this paper uses the bulk method to
compute the turbulent fluxes, which requires temper-
ature, humidity and windspeed at the 2 m level. This
means that it is important to make a distinction be-
tween these variables in the free atmosphere and in
the katabatic surface layer. Most importantly,  dif-
fers from , such that the sensitivity


 is
smaller than 1 (Table 1). If  were not explicitly cal-
culated but simply set equal to  (


=1),
using the bulk method would result in a value of 
that is too high. The over-estimation of  in such
a case is large: about 70%, which is much larger than
Greuell and Böhm (1998) found for the Pasterze in the
Austrian Alps (22%). Other mass balance models that
are based upon a calculation of the energy balance of-
ten use a simplification of the bulk method (e.g. Oer-
lemans, 1992, Van de Wal and Oerlemans, 1994):
 
  (9)
where  is the transfer coefficient,  is the air tem-
perature and  is the surface temperature. These
models use the temperature in the free atmosphere
() for  in equation 9 and do not require .
For  

  we find values of 3.5–5
JÖKULL No. 52, 2003 15