 |
| Titel |
A framework for identifying tailored subsets of climate projections for impact and adaptation studies |
| VerfasserIn |
Jean-Philippe Vidal, Benoit Hingray |
| Konferenz |
EGU General Assembly 2014
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 16 (2014) |
| Datensatznummer |
250093269
|
| Publikation (Nr.) |
 EGU/EGU2014-7851.pdf |
|
|
|
|
|
| Zusammenfassung |
| In order to better understand the uncertainties in the climate of the next decades,
an increasingly large number of increasingly diverse climate projections is being
produced by the climate research community through coordinated initiatives (e.g.,
CMIP5, CORDEX), but also through more specific experiments at both the global
scale (perturbed parameter ensembles) and the regional-to-local scale (empirical
statistical downscaling ensembles). When significant efforts are put into making such
projections available online, very few works focus on how to make such an enormous
amount of information actually usable by the impact and adaptation community.
Climate services should therefore include guidelines and recommendations for
identifying subsets of climate projections that would have (1) a size manageable by
downstream modelling approaches and (2) the relevant properties for informing adaptation
strategies.
This works proposes a generic framework for identifying tailored subsets of
climate projections that would meet both the objectives and the constraints of a
specific impact / adaptation study in a typical top-down approach. This decision
framework builds on two main preliminary tasks that lead to critical choices in
the selection strategy: (1) understanding the requirements of the specific impact /
adaptation study, and (2) characterizing the (downscaled) climate projections dataset
available.
An impact / adaptation study has two types of requirements. First, the study may aim at
various outcomes for a given climate-related feature: the best estimate of the future, the range
of possible futures, a set of representative futures, or a statistically interpretable ensemble of
futures. Second, impact models may come with specific constraints on climate
input variables, like spatio-temporal and between-variables coherence. Additionally,
when concurrent impact models are used, the most restrictive constraints have to be
considered in order to be able to assess the uncertainty associated from this modelling
step.
Besides, the climate projection dataset available for a given study has several
characteristics that will heavily condition the type of conclusions that can be reached. Indeed,
the dataset at hand may or not sample different types of uncertainty (socio-economic,
structural, parametric, along with internal variability). Moreover, these types are present at
different steps in the well-known cascade of uncertainty, from the emission / concentration
scenarios and the global climate to the regional-to-local climate.
Critical choices for the selection are therefore conditioned on all features above. The type
of selection (picking out, culling, or statistical sampling) is closely related to the study
objectives and the uncertainty types present in the dataset. Moreover, grounds for picking out
or culling projections may stem from global, regional or feature-specific present-day
performance, representativeness, or covered range.
An example use of this framework is a hierarchical selection for 3 classes of impact
models among 3000 transient climate projections from different runs of 4 GCMs, statistically
downscaled by 3 probabilistic methods, and made available for an integrated water resource
adaptation study in the Durance catchment (southern French Alps). This work is part of the GICC
R2D2-20501 project
(Risk, water Resources and sustainable Development of the Durance catchment in 2050) and the EU FP7
COMPLEX2
project (Knowledge Based Climate Mitigation Systems for a Low Carbon Economy). |
|
|
|
|
|
|