Geoscience Reference
In-Depth Information
Table 1
GCMs included in the
TAR models
AR4 models
analysis
CCCma
BCM2
CSIRO
CGHR
ECHAM4
CNCM3
GFDL99
CSMK3
HADCM3
ECHOG
NIES99
FGOALS
GFCM20
GFCM21
GIAOM
GIER
HADCM3
HADGEM
INCM3
IPCM4
MIHR
MIMR
MPEH5
MRCGCM
NCCCSM
NCPCM
Note that not all models have
data for all scenarios and
parameters investigated
For the analysis of MAE, MBE and d
r
, we use the mean of the CRU TS 3.1 (Harris et al.
2013
) and Willmott & Matsuura 3.1 data sets (
http://climate.geog.udel.edu/*climate
)
asa
benchmark against which to evaluate GCM output. By averaging two separate data sets,
our confidence in the reliability of the precipitation estimates increases, as different data
sets use slightly different approaches to calculate the same estimate.
Climate projections are critical to adaptation planning, for example, for long-term
planning of infrastructure, power production and agriculture. We therefore test whether
these projections agree with the observations of recent deviations with regard to which
Arctic drainage basins will be most affected by climate change. The aim is to investigate
whether a set of highly divergent future climate projections, together with recent obser-
vations of climate deviations, can form a consistent basis for prioritizing monitoring.
Under limited resources, one either has to rely on some degree of certainty in the distri-
bution of future changes and prioritize monitoring based on this distribution, or try to
harmonize the different possible bases for rational monitoring prioritization that may arise
from uncertain change projections.
In order to include as wide a spectrum of future climate change as possible over the next
half-century, we formulate for this analysis two alternate final stages of possible climates.
We select the five warmest or wettest models from the most severe IPCC Special Report on
Emission Scenarios (SRES) scenario (A2), and the five coldest or driest models from the
least severe scenario (B1). For details on the procedure and the selected models, we refer to
Bring and Destouni (
2013
). For these two cases, we analyze the relative distribution of
projected climate change severity across the 14 largest Arctic drainage basins and compare
this distribution with recent observed deviations from the 1961 to 1990 climate. We
emphasize here that, in absolute terms, we do expect divergence between the scenarios,
