Item

Abstract

This paper investigates the impact of different ocean initialization strategies on the forecast skill of decadal
prediction experiments performed with the ECHAM5/Max Planck Institute Ocean Model (MPI-OM) coupled
model. The ocean initializations assimilate three-dimensional temperature and salinity anomalies from two different
ocean state estimates, the ocean reanalysis of the German contribution to Estimating the Circulation and
Climate of theOcean (GECCO) and an ensemble ofMPI-OM ocean experiments forced with the NCEP–NCAR
atmospheric reanalysis. The results show that North Atlantic and Mediterranean sea surface temperature (SST)
variations can be skillfully predicted up to a decade ahead and with greater skill than by both uninitialized simulations
and persistence forecasts. The regional distribution of SST predictive skill is similar in both initialization
approaches; however, higher skill is found for the NCEP hindcasts than for the GECCO hindcasts when a combination
of predictive skill measures is used. Skillful predictions of surface air temperature are obtained over
northwestern Europe, northern Africa, and central-eastern Asia. The North Atlantic subpolar gyre region stands
out as the region with the highest predictive skill beyond the warming trend, in both SST and upper-ocean heatcontent
predictions. Here the NCEP hindcasts deliver the best results due to a more accurate initialization of the
observed variability. The dominantmechanism for North Atlantic climate predictability is of dynamical origin and
can be attributed to the initialization of the Atlantic meridional overturning circulation, thus explaining the reoccurrence
of high predictive skill within the second pentad of the hindcasts experiments. The results herein
demonstrate that ocean experiments forced with the observed history of the atmospheric state constitute a simple
but successful alternative strategy for the initialization of skillful climate predictions over the next decade.