Geoscience Reference
In-Depth Information
DNMI [3]. With the use of GIS software, these two maps were
overlaid, allowing stations in closest proximity to the main
area(s) of agricultural activity in each county to be identified and
selected. This choice was heavily constrained by the availability
of continuous time series data over our period of study (due to
some stations being built after 1958, some being taken out of
service for some years, and others being closed down), and by
the fact that not all weather stations had the facilities to collect
both precipitation and temperature data. In some cases, output
from more than one station was averaged to produce the data set
for a county, for example, where it spanned a broad geographic
area and no single weather station was thought to be solely
representative. In other cases, data from neighbouring coun-
ties were also incorporated, based on the assumption that they
contributed relevant information about the weather conditions,
which stations situated in the county may not have captured due
to their location. Where data were simply unavailable and there
were no suitably placed stations in neighbouring counties to pro-
vide proxy data, the time period in question was omitted from
our analysis for that county.
*
Finally, on three occasions, indi-
vidual observations were interpolated.
†
In these instances, only
1 month's data were missing from an otherwise complete series.
Analysis at the
national level
In order to conduct regression analyses at the national level, it
was necessary to produce aggregate weather and crop figures
based on the county data used in previous analyses. Production
of each crop per county was calculated as a proportion of total
national output (for that crop), and then weather data were
weighted accordingly. This gave weather data in counties pro-
ducing a larger share of the national yield (such as in South-
Eastern Norway) a higher weight than in those counties where
production of that crop was lower. Where data were omitted
from analysis at the county level, it was, by necessity, also
excluded at the national level.
the RegClim
scenario
Projected future values for GDD and annual precipitation were
obtained from the RegClim Project—a regional climate sce-
nario for Northern Europe until 2050 [17]. Regional Climate
Development Under Global Warming Project (RegClim) uses
an 'Atmospheric Regional Climate Model to estimate the
regional climate in Northern Europe and adjacent sea areas,
*
That is, Telemark 1990-2001 and Hedmark 1999-2001.
†
That is, Telemark: precipitation, August 1989; Hedmark: GDD, August 1987
and May 1989.
