Jökull - 01.01.2011, Blaðsíða 1

Reviewed research article
A data set of gridded daily temperature in Iceland, 1949–2010
Philippe Crochet and Tómas Jóhannesson
Icelandic Meteorological Office, Bústaðavegur 9, IS-150 Reykjavík, Iceland
Corresponding author: philippe@vedur.is
Abstract — A high spatial resolution data set of gridded daily surface air temperature in Iceland has been
derived for the period 1949–2010. Between 60–80% of the daily estimates are within ±1 ◦C depending on
the month of the year and between 90–95% are within ±2 ◦C of independent station measurements in 1995–
2010. The data set is thus well suited for various hydrological, glaciological and climatological modelling
studies. The quality of the gridded data set is found to be best near the coast because of the higher station
density. Including data from automatic stations, the establishment of which started in the early 90’s, was found
to improve the data set. Derived 30-year mean monthly maps compare favourably with reference maps derived
directly from monthly mean station temperatures with more sophisticated statistical techniques. An analysis of
decadal temperature variations based on the data set shows that the decade 2001–2010 was the warmest of the
last 60 years and makes it possible to identify spatial patterns in the decadal variations. As an example, the
spatial distribution of the warming of recent decades shows that it is more pronounced in the inland compared
with coastal areas.
INTRODUCTION
Knowledge of the spatio-temporal distribution of sur-
face air temperature and precipitation in the complex
terrain of Iceland is important for various applica-
tions ranging from local and regional climate monitor-
ing to hydrological and glaciological modelling stud-
ies at the catchment scale. Gridded data sets with a
high temporal resolution and a good spatial coverage
are particularly suitable for modelling because they
bridge gaps in the station network and do not depend
directly on the operation of observing stations at vari-
ous locations.
Glacier mass balance studies in Iceland are of-
ten carried out with simple temperature-index melt
models using extrapolated temperature and precipi-
tation from single station measurements employing
fixed horizontal and vertical temperature and precipi-
tation gradients (Jóhannesson et al., 1995; Jóhannes-
son, 1997; Aðalgeirsdóttir et al., 2006). The accu-
racy of this type of modelling could be improved with
accurate gridded temperature and precipitation esti-
mates, in particular for mass balance modelling of the
8100 km2 Vatnajökull ice cap where the single station
approach is clearly inadequate. Gridded data sets are
also useful for mass balance modelling of glaciers that
are far away from meteorological stations and where
station measurements are missing for single days or
longer periods due to instrument failures or other rea-
sons.
Hydro-glaciological models that simulate the re-
sponse of small to medium-sized river catchments in
complex terrain often need to operate at high spatio-
temporal resolutions, typically 1 km and 1 day, be-
cause the variability of precipitation and temperature
is large and response times are short. Snow and
glaciers in particular play an important role in the
flow regime characteristics of many Icelandic rivers
and runoff generation depends to a great extent on
the interplay of rain and the melting of snow and ice
(Jónsdóttir et al., 2008). Under these conditions, it
JÖKULL No. 61, 2011 1