Jökull - 01.01.2005, Qupperneq 135

Energy balance of Brúarjökull and the August 2004 floods in the river Jökla.
river discharge of Jökla (550 m3s−1) could be
explained by glacial melting, 60% of which was
due to the net radiation and 40% from the high
turbulent heat fluxes (Figure 12a).
2. The first flood in Jökla, August 3–6, when
only ∼65% of the average river discharge (750
m3s−1) could be explained by glacial melting
(Figure 12a); 72% of the melting was con-
tributed by the net radiation and 28% by the tur-
bulent heat fluxes. Thus,∼35% of the river dis-
charge was due to the intensive rainfall during
August 1-3 (Table 1).
3. The second flood in Jökla, August 9-14, when
the average river discharge (690 m3s−1) could
be fully explained by the net energy supplied
for glacial melting (Figure 12a). The exception-
ally high temperatures and stable winds of∼4.5
m s−1 maintained high turbulent heat fluxes,
and the high solar radiation, along with abruptly
reduced albedo, led to strong net radiation (Fig-
ure 12). The contribution from the net radiation
was 69% while 31% came from the turbulent
heat fluxes.
Table 1: Accumulated precipitation at Akurnes dur-
ing August 1–4, 2004 (see Figure 1 for location). –
Úrkoma á Akurnesi dagana 1.–4. ágúst 2004.
Time interval Ps
dd.mm hh:mm mm of water
31.07 09:00 – 31.07 18:00 0
31.07 18:00 – 01.08 09:00 0.5
01.08 09:00 – 01.08 18:00 21.5
01.08 18:00 – 02.08 09:00 25.5
02.08 09:00 – 02.08 18:00 2
02.08 18:00 – 03.08 09:00 64.7
03.08 09:00 – 03.08 18:00 12
03.08 18:00 – 04.08 09:00 0
Runoff estimated from empirical models
Three empirical ablation models were considered re-
lating the daily ablation (as in mm of water) at an ele-
vation hG on the glacier to daily mean air temperature
TS at an elevation hS , 20 km away from the glacier
front (at Eyjabakkar 655 m a.s.l., see Figure 1):
as =
{
ddf · T̂ ; T̂ ≥ 0
0; T̂ < 0
, (4)
(5)
as =
{
(MF+ a · Q)T̂ ; T̂ ≥ 0
0; T̂ < 0
,
(6)
as =
⎧
⎨
⎩
TF · T̂ + b · Q
ρL
; (TF · T̂ + b · Q) ≥ 0
0; (TF · T̂ + b · Q) < 0
,
where T̂ = (TS + γ(hG − hS)). The term γ =
−0.6×10−2 ◦Cm−1 is a constant lapse rate andQ is a
theoretically calculated clear-sky irradiance (Olseth et
al., 1995). Eq. (4) is a degree-day model (DDM) (e.g.
Jóhannesson et al., 1995), Eq. (5) is the temperature-
index model (TIM) introduced by Hock (1999) and
Eq. (6) is an empirical energy balance model (EEB).
The scaling factors, MF and TF, are assumed to be
constant and ddf , a and b to have one value for snow
and another for firn/ice (Table 2). The model in Eq.
(5) is the same as given by Gudmundsson et al. (2003,
p. 5–6) except that the long-wave radiation balance is
ignored. TF is a turbulent heat flux scaling factor and
b is mainly reflecting the surface albedo (on average
for snow or firn/ice) and the average cloud cover. The
term ρL is the same as in Eq. (2). The advantage of
Eqs. (4–6) is the use of air temperature, observed at
non-glaciated area away from the glacier front, as the
only input.
The coefficients of the empirical models (Table 2)
were optimized for Brúarjökull by using the energy
balance calculations at the AWSs (Eq. 2) from 1996-
2001, and then tested for the years 2002-2004. Glacial
runoff towards Jökla calculated by Eq. (3) with a s
from Eqs. (4-6) (DDDM , DTIM , andDEEB , respec-
tively), was compared to glacial runoff (DM ) derived
from the energy budget (Figure 13). Of the three mod-
els, the EEB model in Eq. (6) gave the best prediction
of the total glacial runoff; it is the only model that
does not underestimate the melting before mid-July,
JÖKULL No. 55 135