caa a22 4500 605474850 CHVBK 20210128100349.0 cr unu---uuuuu 210128e20150501xx s 000 0 eng 10.1007/s00382-014-2370-7 doi (NATIONALLICENCE)springer-10.1007/s00382-014-2370-7 Added-value from initialization in predictions of Atlantic multi-decadal variability [Elektronische Daten] [J. García-Serrano, V. Guemas, F. Doblas-Reyes] Identifying regions sensitive to external radiative changes, including anthropogenic (sulphate aerosols and greenhouse gases) and natural (volcanoes and solar variations) forcings, is important to formulate actionable information at multi-year time-scales. Internally-generated climate variability can overcome this radiative forcing, especially at regional level, so that detecting the areas for this potential dominance is likewise critical for decadal prediction. This work aims to clarify where each contribution has the largest effect on North Atlantic sea surface temperature (SST) predictions in relation to the Atlantic multi-decadal variability (AMV). Initialized decadal hindcasts and radiatively-forced historical simulations from the fifth phase of the Climate Model Intercomparison Project are analysed to assess multi-year skill of the AMV. The initialized hindcasts reproduce better the phase of the AMV index fluctuations. The radiatively-forced component consists of a residual positive trend, although its identification is ambiguous. Initialization reduces the inter-model spread when estimating the level of AMV skill, thus reducing its uncertainty. Our results show a skilful performance of the initialized hindcasts in capturing the AMV-related SST anomalies over the subpolar gyre and Labrador Sea regions, as well as in the eastern subtropical basin, and the inability of the radiatively-forced historical runs to simulate the horseshoe-like AMV signature over the North Atlantic. Initialization outperforms empirical predictions based on persistence beyond 1-4years ahead, suggesting that ocean dynamics play a role in the AMV predictability beyond the thermal inertia. The initialized hindcasts are also more skilful at reproducing the observed AMV teleconnection to the West African monsoon. The impact of the start date frequency is also described, showing that the standard of 5-year interval between start dates yields the main features of the AMV skill that are robustly detected in hindcasts with yearly start date sampling. This work updates previous studies, complementing them, and concludes that skilful initialized multi-model forecasts of the AMV-related climate variability can be formulated, improving uninitialized projections, until 3-6years ahead. Springer-Verlag Berlin Heidelberg, 2014 Decadal climate prediction nationallicence Atlantic multi-decadal variability nationallicence Multi-model ensemble hindcasts nationallicence García-Serrano J. Institut Català de Ciències del Clima (IC3), Barcelona, Spain aut Guemas V. Institut Català de Ciències del Clima (IC3), Barcelona, Spain aut Doblas-Reyes F. Institut Català de Ciències del Clima (IC3), Barcelona, Spain aut Climate Dynamics Springer Berlin Heidelberg 44/9-10(2015-05-01), 2539-2555 0930-7575 44:9-10<2539 2015 44 382 https://doi.org/10.1007/s00382-014-2370-7 text/html Onlinezugriff via DOI BK010053 XK010053 XK010000 Metadata rights reserved Springer special CC-BY-NC licence nationallicence 1 research-article jats NATIONALLICENCE NATIONALLICENCE NL-springer NATIONALLICENCE 856 40 https://doi.org/10.1007/s00382-014-2370-7 text/html Onlinezugriff via DOI NATIONALLICENCE 700 1- García-Serrano J. Institut Català de Ciències del Clima (IC3), Barcelona, Spain aut NATIONALLICENCE 700 1- Guemas V. Institut Català de Ciències del Clima (IC3), Barcelona, Spain aut NATIONALLICENCE 700 1- Doblas-Reyes F. Institut Català de Ciències del Clima (IC3), Barcelona, Spain aut NATIONALLICENCE 773 0- Climate Dynamics Springer Berlin Heidelberg 44/9-10(2015-05-01), 2539-2555 0930-7575 44:9-10<2539 2015 44 382