ARCTIC, ANTARCTIC, AND ALPINE RESEARCH © 3
Iceland. Specifically, we answer the following research
questions: First, what are the components of a regional
climate model for Iceland specific to fisheries and aqua-
culture? Second, what broad socioeconomic aspects
related to informants’ experiences and perceptions of
climate impacts and threats should be considered in
future climate adaptation research on Icelandic fisheries
and aquaculture?
Methods
Climate modeling and related analysis
To assess potential changes of weather conditions rele-
vant for fishing activities, we analyzed a single-model
initial-condition large ensemble, the SMHI Large
Ensemble (SMHI-LENS; Wyser et al. 2021). SMHI-
LENS employs the general circulation model EC-
Earth3.3.1 (Döscher et al. 2022), a state-of-the-art global
climate model that was used for a wide range of model
experiments in the Sixth Phase of the Coupled Model
Intercomparison Project (CMIP6; Eyring et al. 2016)
and related efforts. EC-Earth3.3.1 makes use of an atmo-
spheric model component, based on the Integrated
Forecast System (cycle 36r4) of the European Center
for Medium-Range Weather Forecasts coupled to
NEMO3.6 for the ocean (Rousset et al. 2015) and
LIM3 for sea ice (Rousset et al. 2015). For SMHI-
LENS, EC-Earth3.3.1 is integrated with a horizontal
resolution of approximately 80 km in the atmosphere
and a tripolar 1° grid in the ocean, the latter featuring
grid refinement of 1/3° near the Equator. In the vertical,
the ocean component operates with seventy-five layers,
and the atmospheric model features ninety-one levels
from the Earth’s surface up to a model top at 1 Pa.
SMHI-LENS consists of fifty ensemble members,
each individual simulation starting from different initial
conditions but all following identical external forcings,
such as greenhouse gas concentrations, aerosols, solar
variability, etc. One of the main advantages of such
a single-model large ensemble is isolating the forced
response and the internal variability in the coupled
climate system (Maher, Milinski, and Ludwig 2021).
Though each individual simulation features its own
phase of internal variability, they all show a response
to the identical external forcing, so that analysis of, for
example, the ensemble mean results in averaging out the
“noise” of internal (natural) variability. The simulation
period of SMHI-LENS covers the recent historical per
iod (1970-2014) —for which observed external forcings
in line with CMIP6 recommendations are applied— and
‘he remainder of the twenty-first century (2015-2100)
following the forcing specifications given in the
ScenarioMIP protocol (O’Neill et al. 2018). SMHI-
LENS was produced employing a wide range of different
juture scenarios (currently eight different scenarios are
publicly available). For this study, we present analysis of
Shared Socioeconomic Pathway (SSP) 3-7.0, which is
a scenario marked by a continuous increase of green-
house gas emissions throughout the twenty-first cen-
‚ury, resulting in an atmospheric CO, concentration of
above 800 ppm in 2100, which is associated with an
anthropogenically induced radiative forcing of approxi-
mately 7 W m” and a relatively strong global mean
temperature increase of roughly 4°C compared to pre-
industrial levels. We chose S$S$P3-7.0 because one aim of
‘he research was to present an example of potential
uture wind changes, for which it is more illustrative to
ıse a higher emission scenario, and SS$SP3-7.0 belongs to
he Tier 1 scenarios of CMIP6. However, there are
a range of other CMIP6-Tier 1 emission and land use
scenarios in existence that would lead to smaller global
mean temperature increases (e.g., SSP1-2.6, SSP2-4.5,
5S$P4-6.0) or to an even higher temperature increase
(SSP5-8.5).
From the exploratory interviews (described below), it
was common knowledge that a mix of climatic condi-
ions including wind speed and direction, wave height
and period, as well as sea and air temperature affect
fishing activities at sea, and therefore it can be difficult
oO set accurate thresholds in which these factors cause
disruptions. Using rough thresholds identified by key
'nformants, the decision was made to focus on wind
speed as the primary variable of weather and sea condi-
ions in which fishing could not be safely carried out.
The absolute values of wind speeds output by climate
models cannot be directly compared to observed wind
speeds. Climate models solve the underlying physical
equations only for representative points in a rather
coarse grid (approximately 80 km in our case) and
:ypically feature biases in absolute wind speeds (and
other variables). Thus, we decided to use thresholds
based on the climatology of the model itself to define
a proxy for storm days on which fishing is not safe. This
proxy threshold is marked by the ninety-fifth percentile
of daily wind speeds of the fifty members of the histor-
:cal data experiment (reference period 1970-2014), cal-
culated individually for every grid box of the climate
nodel (see Appendix Figure A1 including a comparison
:©O the equivalent ninety-fifth percentile for the observa-
ion-based ERA5 reanalysis; Hersbach et al. 2020).
Therefore, we define thresholds matching wind speeds
‘hat are exceeded on average 18 days a year (5 percent of
365 days). For the future scenarios we keep these thresh-
olds and analyze whether exceedances occur more or
‚ess frequently in the climate model,
