CERFACS CERFACS Scientific Activity Report Jan. 2010 – Dec. 2011

CLIMATE VARIABILITY AND PREDICTABILITY
potentially lead to decadal signals over the adjacent continents like in observations. The realism of the
simulated modes has been first quantified. The mechanisms at the origin of their excitation have been then
investigated based on classical statistical lead-lag analysis. We show that in CNRM-CM5, a maximum
of AMOC is preceeded, about 30-yr before, by an anomalous cyclonic circulation off Europe leading to
enhanced meridional ocean heat transport into the subpolar gyre as well as into the GIN seas. Excess of
heat there induces a strong sea-ice response leading to salinity anomalies that are advected into the subpolar
gyre through the Denmark Strait, thus further reinforcing the direct acceleration of the subpolar gyre due
to overlying atmospheric forcing. This dominant mode for AMOC is damped by the northward advection
of subtropical fresher water along the subtropical gyre. Predictability studies within the so-called “prefect
model framework” has been started for the AMO and will be completed in 2012.
2.2.4 The decadal signature of sea-level changes
The ANR CECILE project started in 2010. It aims at evaluating the impact of sea level variations, including
future sea level rise, on erosion and the coast lines. In collaboration with CNRM, Cerfacs provides expertise
and decadal forecast simulations to make this evaluation both retrospectively and prospectively. The work
done during this first phase of the project was directed towards evaluation of the different experiments
produced so far. First comparison to the reanalyses, including NEMOVAR, have been compared to quality
controled, long tide gauge time series and sea level reconstructions. Second, a first order predictability
evaluation of the sea level regional anomalous patterns in both retrospective decadal hindcasts, hereafter
DEC, and historical XXth century simulations, hereafter HIST, has been conducted using the reliability
approach. First results show that the DEC ensemble spatial spread is smaller than HIST which is itself
underestimated compared to NEMOVAR. Second, preliminary analyses suggest that the predictability at
regional timescale in both DEC and HIST ensemble can barely be demonstrated using single model 10-
member ensemble hindcasts.
120 Jan. 2010 – Dec. 2011