Jökull - 01.01.2013, Blaðsíða 98

H. Ágústsson et al.
Comparison between the observed winter precip-
itation at the two lowland weather stations and the
winter balance at the survey sites on Mýrdalsjökull
reveals a correlation (Figure 4) between observations
from individual stations and sites. In light of the con-
siderable variability of the data and that a robust re-
lationship is not to be expected, we assume a sim-
ple linear relationship and make no assumption on
the offset. On average, the measured winter bal-
ance at sites M1 (greatest) and M3 (lowest) is two
and three times greater than the observed precipitation
at Vík (mean ratios 3.1 and 2.1, respectively), with
the corresponding ratio (2.4) at M2 lying in between
the other two. The smaller precipitation observed at
Vatnsskarðshólar leads to a larger ratio between the
winter balance and the observed precipitation (4.8 for
M1, 3.8 for M2 and 3.3 for M3). Although the ratio
of precipitation between the lowlands and the ice cap
plateau is not necessarily similar during summer and
winter (Rögnvaldsson et al., 2007), the above ratios
may be used to give a first estimate of the precipitation
on the ice cap plateau during summer, i.e. between
the spring and autumn measurements (Guðmundsson,
2000). With the current lack of data, this estimate is
important and of relevance, e.g. in the context of cal-
culating summer runoff.
Table 3. Mean observed precipitation [m] at Vík
in Mýrdalur and Vatnsskarðshólar and estimated
precipitation [m] at sites on Mýrdalsjökull during
summer (May–Aug.) 2007–2011, as well as esti-
mated mean annual precipitation [m] (Sept.–Aug.)
2007–2011. – Meðalúrkoma [m] í Vík í Mýrdal
og á Vatnsskarðshólum og áætluð sumarúrkoma á
mælistöðum á Mýrdalsjökli (maí–ágúst) 2007–2011,
auk áætlaðar meðalársúrkoma [m] á mælistöðum á
jökli (sept.–ágúst) 2007–2011.
Obs. M1 M2 M3
Summer Vík 0.58 1.8 1.4 1.2
Vat. 0.31 1.5 1.2 1.0
Ratio 1.87 1.20 1.17 1.20
Annual Vík 2.60 8.1 6.4 5.6
Vat. 1.62 7.8 6.1 5.4
For May–August of 2007–2011 the (summer) pre-
cipitation at Vík (0.49–0.77 m) and Vatnsskarðshólar
(0.23–0.42 m) is respectively 28% and 24% of the
mean observed winter balance at the stations, i.e. for
Sept.–April. Based on this ratio, the survey sites on
the glacier should see 1–1.8 m of precipitation during
the summer (Table 3). The estimated summer pre-
cipitation would be approx. 20% lower when based
on observed precipitation at Vatnsskarðshólar than at
Vík. On an annual basis the estimated precipitation at
the sites, i.e. measured winter balance plus estimated
summer precipitation, ranges from 5.4 m of water at
M3 to 8.1 m at M1.
SIMULATED PRECIPITATION
In the RÁV-project (Reikningar Á Veðri; Rögnvalds-
son et al., 2011), weather in Iceland has been dynami-
cally downscaled using the non-hydrostatic mesoscale
Advanced Research WRF-model (ARW, Skamarock
et al., 2005). This state of the art numerical atmo-
spheric model is used extensively both in research
and in operational weather forecasting throughout the
world, including Iceland. The atmospheric modeling
is done at high resolution, 9 and 3 km in the horizon-
tal and 55 levels in the vertical. The model is forced
by atmospheric analysis from the European Centre for
Medium-Range Weather Forecasts (ECMWF). The
model takes full account of atmospheric physics and
dynamics, and the relevant parameterization scheme
for this study is the moisture scheme of Thompson
et al. (2004); other details of the setup of the model
are found in Rögnvaldsson et al. (2011). One of the
key aspects of the dataset is its high spatial resolution,
but as resolution is increased, the atmospheric flow
and its interaction with the complex orography are in
general better reproduced. In short, the RÁV-dataset
is currently the most accurate and detailed dataset de-
scribing the state of the atmosphere above Iceland,
at high temporal and spatial resolutions in 4 dimen-
sions. At a horizontal resolution of 3 km, the large
scale features of the orography of Mýrdalsjökull are
adequately described in the atmospheric model; the
elevation of sites M1 and M2 is correct while at M3,
the ice cap’s maximum elevation is underestimated by
approx. 100 m in the model (cf. Figures 1 and 5).
98 JÖKULL No. 63, 2013