Geoscience Reference
In-Depth Information
Considerable advances have been made between the two successive generations of
model ensemble runs that underlie the latest available full IPCC Assessment Reports, the
TAR and the AR4. Examples include improved parameterization of land surface schemes
and snowpack, as well as the inclusion of canopy processes in most models (Randall et al.
2007
). However, remaining uncertainties include the representation of cryospheric feed-
backs, which explain part of the range of model responses at mid- and high latitudes
(Randall et al.
2007
).
These improvements are suggested by the fact that the results indicate a lower spread
among the ensemble members for the AR4 models and, in the case of precipitation, a
marked improvement in the agreement with absolute observed values. However, despite
the improvement in precision, accuracy remains insufficient for precipitation projections,
which, even if they were correct, are still difficult to translate to subsequent changes in the
water cycle without further and finer resolved hydrological modeling. The uncertainty in
observations, particularly gauge undercatch of solid precipitation due to wind (e.g., Yang
et al.
2005
; Tian et al.
2007
), may explain part of the remaining difference between
observations and the improved AR4 results.
The large span in model performance, evident from the explicit error and performance
analysis of AR4 models, also indicates that large uncertainties and shortcomings remain for
reliable simulations of hydro-climatic parameters on basin scales. The analysis furthermore
underlines that models may yield good output results on these scales for the wrong reasons.
For example, the pan-Arctic bias error (MBE) of temperature for the GIER model is close
to zero. This would place it at the top of a basin-scale simulation performance ranking
based on this measure. At the same time, the absolute error (MAE) and the performance
index (d
r
) for the same model are among the worst of the ensemble, implying large
deviations from observations at individual grid points, even though the deviations happen
to almost cancel out across the PADB.
The results for climate deviation and change severity across different basins further
indicate that the relative distribution of climate change simulated by AR4 GCM projections
does not agree with the relative distribution of currently observed climate deviations across
the major Arctic basins. This complicates the choice of a robust prioritization strategy for
hydrological monitoring development based on the reconciliation of observations and
projections for the assessment of which Arctic river basins that are/will be most affected by
climate change; this complication is further discussed in Sect.
6.3
.
6.2 Arctic Hydro-Climatic Change
In the Arctic, certain counterintuitive hydro-climatic changes have caused extensive
investigation and discussion in the scientific community. The fact that most of the 13 major
rivers we study can be termed so-called excess rivers (i.e., the increase in discharge has
been greater than the increase in precipitation) is consistent with findings by Milliman et al.
(
2008
) for a related set of Arctic rivers. This and other studies (e.g., Dyurgerov and Carter
2004
; McClelland et al.
2006
; Smith et al.
2007
; Lyon et al.
2009
; Brutsaert and Hiyama
2012
) indicate that both specific climatic and subsurface processes pertaining to the Arctic,
such as permafrost degradation, as well as other anthropogenic changes and general
atmospheric patterns, including increased moisture transport from outside the PADB, are
responsible for the range in discharge patterns. Changes in glacier mass balance cannot
explain the results of increasing river flows as the vast majority of Arctic glacier area and
volume is located outside the major river basins (Dyurgerov et al.
2010
). However, the
relative importance of the different contributing factors is not definitely established.
