The workshop on Climate Prediction in the Atlantic-Arctic sector was jointly organised by the Bjerknes Climate Prediction Unit and the EU Climate Modelling Cluster on 5-7 June 2019 in Bergen (NO) and is considered the 2nd EU Modelling Cluster workshop.

The EU Climate Modelling Cluster collects a number of H2020 projects working on modelling topics such as Blue-Action, APPLICATE, PRIMAVERA and CRESCENDO and the BCPU project, funded by the Trond Mohn Foundation. The cluster was initiated by the EC/EASME in May 2017 with a workshop on evaluating climate and Earth system models at the process level.


Key take-home messages for the scientific community

  • Climate prediction in the Atlantic and Arctic sector is reaching useful skill levels. There appears to be great potential to enhance prediction skill in our region, as recent results indicate that current climate prediction systems underestimate the predictable component of the climate system. In particular, the high predictability of the ocean in these systems does not transfer to atmospheric circulation to the extent expected, and predictability over continents appears too low. Low forecast skill is not completely surprising, given the large morel errors, discrepancies among initial conditions in different forecast systems (especially in the large-scale ocean circulation), and initial forecast shock and drift. Although there is a general understanding of the drivers of large-scale seasonal to decadal predictability (e.g., El Niño Southern Oscillation, Atlantic multi-decadal variability), the key physical processes are still not well understood and this hampers further improvement in forecast skill.
  • In the short term, forecast skill can be enhanced by using large (40 members or more) single and multi-model ensembles to identify the predictable signal from chaotic atmospheric noise, and by combined dynamical-empirical approaches. At present increasing ensemble size is likely more beneficial than increasing resolution, which can lead to reduced model biases but not necessarily to enhanced prediction skill. Large-ensembles are also critical for predicting extreme event probabilities. Skill improvement can also be achieved through the development of methods to combine high-resolution and low-resolution ensembles, optimally utilize multi-model ensembles, and approaches to blend dynamical and machine-learned ensembles.
  • In the medium term, improved forecast skill may be achieved through the application of advanced data assimilation approaches that can make effective use of sparse data and can consistently initialize the different components of the climate system. There is also a need to develop appropriate methods to generate ensembles that sample key uncertainties in the various components of the climate system. Forecast applications often require bias correction approaches, although such corrections introduce uncertainty and can reduce skill. Refinement of a posteriori bias correction approaches to consider higher order moments is required. Alternative to standard dynamical bias correction can also be beneficial, as flux correction approaches are often not useful.
  • In the long-term, improvements in understanding of key physical processes are necessary to reduce model error and to improve forecasts. The atmospheric boundary layer in particular (where many parameterizations are active) suffers from a number of biases that need to be reduced, ranging from the polar latitudes to the tropics and from continental areas to the oceans.
  • Observations are critical to improve models and predictions, as well as to monitor climate.  Extending together the OSNAP and RAPID measurements is important to understand how subpolar and subtropical Atlantic Meridional Overturning Circulation (AMOC) variability are connected, while measurements of the deep in the deep ocean including the overflows are need to understand the chain of events leading to multi-decadal ocean variability over the region. The development of higher resolution models demands new observations that can resolve the simulated phenomena, this requires both observational campaigns and long-term observational programmes, and for greater communication between modellers and observers. Data assimilation approaches can be used to identify key regions and variables to observe, and to provide observational uncertainty estimates.
  • Enhancing model resolution is a promising avenue, but eddy and convective resolving resolutions required to make stepwise improvement are beyond the reach of current computational resources. There are several major challenges facing the development of the next generation of high-resolution climate models: we need to develop strategies to use the new computer infrastructure (e.g., hybrid cpu,gpu) that are rapidly evolving and not designed for climate (weather) modelling purposes, we need an appropriately trained workforce, we need to adopt community models, while still allowing an appropriate level of model diversity. Data from high resolution models and large ensembles will need to be shared to maximize knowledge gain, and this will require community infrastructure.


Mechanisms giving rise to climate predictability

Chair: Helene Langehaug, NERSC

Rong Zhang (GFDL, keynote) Mechanisms for decadal climate predictability in the Atlantic-Arctic sector, available in Zenodo

Nour-Eddine Omrani (UiB, BCPU) Understanding the multidecadal Northern Hemisphere climate variably from the perspective of damped Coupled stratosphere/troposphere/Ocean oscillation

Jennifer Mecking (University of Southampton, Blue-Action) Ocean versus Atmosphere in the Eastern North Atlantic Subpolar Gyre Ocean Heat Content

Pablo Ortega (BSC, APPLICATE) A multi-model comparison of the ocean contributions to multidecadal variability in the North Atlantic,

Chair: Fei Li, NILU

Shuting Yang (DMI, Blue-Action, EUCP) On the climate variability and the recent abrupt cooling over Subpolar North Atlantic, available in Zenodo

Jeremy Grist (NOC, Blue-Action, PRIMAVERA) Re-emergence of North Atlantic subsurface ocean temperature anomalies in a seasonal forecast system,

Hilla Gerstman (ETH Zurich, Blue-Action) Stratospheric influence on extreme weather events in the North Atlantic basin

Guillaume Gastineau (SU, Blue-Action) Atmospheric response to the observed sea-ice variability: role of continental snow cover and decadal SST variability,

Chair: Jennifer Mecking, University of Southampton

Johann Jungclaus (MPI, PRIMAVERA, invited) Detecting changes in North Atlantic variability under global warming,

Marius Årthun (UiB, BCPU, Blue-Action) The role of Atlantic heat transport in future Arctic winter sea ice loss,

Paul Kushner (University of Toronto) Fast and slow response of West African precipitation to aerosol forcing,

Chair: Stefan Sobolowski, NORCE

Elisa Manzini (MPI, Blue-Action) Nonlinear Response of the Stratosphere and the North Atlantic-European Climate to Global Warming,

Luigi Vidale (University of Reading, PRIMAVERA) Global Climate Modelling at High Resolution in PRIMAVERA/HighResMIP

Pier Dmitry Sein (AWI, PRIMAVERA) Simulating the Arctic climate with the AWI climate models: From global to regional scales

Challenges to developing climate services

Chair: Tor Eldevik, UiB

Beatriz Balino (UiB, CORA) Brief overview of the Joint Coordination Office for WCRP Regional Activities,

Francisco J. Doblas Reyes (BSC, EUCP, keynote) Transitioning climate prediction from research to operations and services

Anne Britt Sandø (Institute of Marine Research, BCPU) Potential applications of climate predictions on different levels in the marine ecosystem

Mette Skern-Mauritzen (Institute of Marine Research) The use of climate predictions to inform fisheries and ecosystem management – an ICES perspective

Climate predictability limits

Chair: Ingo Bethke, UiB

Jon Robson (University of Reading, invited) Recent multivariate changes in the North Atlantic climate system, with a focus on 2005–2016

Thomas Jung (AWI, APPLICATE) Advanced prediction in polar regions and beyond (APPLICATE): Recent progress

Iuliia Polkova (Universität Hamburg, Blue-Action) Preconditions for cold air outbreaks and prediction skill,;

Helene R. Langehaug (NERSC, BCPU, Blue-Action) Assessing poleward propagation of temperature anomalies in decadal hindcast experiments,

Juliette Mignot (IPSL, Blue-Action, EUCP) IPSL-EPOC decadal prediction system: an update from the trenches,

Daniela Matei (MPI, Blue-Action) Decadal-scale predictive skill of the North Atlantic upper-ocean salt content and its attribution to the initialization of the North Atlantic Ocean Circulation,

Chair: Martin King, NORCE

Rosemary Eade (Met Office, invited) Decadal Variability and Trends with a focus on the North Atlantic Oscillation,

Panos Athanasiadis (CMCC, Blue-Action, PRIMAVERA) Decadal predictability of North Atlantic blocking and the NAO,

Francois Counillon (NERSC, BCPU, Blue-Action) The role of model bias for prediction skill and methods to constrain it,

Data assimilation for reanalysis and model initialization

Chair: Madlen Kimmritz, NERSC

Eugenia Kalnay (UMD, keynote) Two new ways to improve reanalysis: Use future observations to improve the analyses and forecasts, and minimize reanalysis "jumps" when introducing new observing systems

Steve Penny (University of Maryland, invited) Transitioning to strongly coupled data assimilation for Earth system initialization

Benjamin Menetrier (IRIT, invited) Localization for ensemble DA: objective diagnostic and efficient application,

Patrick Laloyaux (ECMWF, invited) Application of coupled data assimilation at ECMWF,

Chair: Francine Schevenhoven, UiB

Yiguo Wang (NERSC, BCPU, Blue-Action) Development of ensemble-based data assimilation techniques for climate prediction,

Victor Estella Perez (LOCEAN, Blue-Action) Reconstructions of the AMOC in the historical period using surface data with the IPSL coupled model,

Madlen Kimmritz (NERSC; BCPU, Blue-Action) The role of ocean and sea ice for seasonal prediction in the Arctic

Filippa Fransner (UiB, BCPU) Ocean biogeochemical predictions - the role of initial conditions and sources of potential predictability

Panel discussion

Franz Immler, Head of Sector Climate Action, EASME, European Commission

Francisco J. Doblas Reyes, Director of Earth Sciences Department at BSC

Erik Kolstad, senior researcher at Regional Climate & Climate Services group, NORCE, and adjunct professor at Centre for Climate and Energy Transformation, UiB

Mette Skern-Mauritzen, Leader of the Ecosystem Processes research group atHavforskningsinstitutt

Tor Eldevik*, Co-leader of the Bjerknes Climate Prediction Unit and Deputy director of the Bjerknes Centre for Climate Research* panel discussion facilitator

Output: Notes available in Zenodo, in Deliverable D6.2



Ramon Fuentes-Franco (SMHI, PRIMAVERA, EUCP) Possible tropical sources of predictability for inter- annual variability of summer precipitation over Nordic European countries,

Hjálmar Hátún (Faroe Marine Research Institute) An inflated subpolar gyre blows life towards the northeastern Atlantic

Valerio Lembo (University of Hamburg, Blue-Action) Prediction of the long-term climate response in a coupled climate model using response theory

 J. Oelsmann (presented by Johann Jungclaus) (MPI-Met) AMOC-related SST variations as a driver of the Atlantic Multidecadal Variability in MPI-ESM1.2

Lea Svendsen (UiB, BCCR, BCPU) Pacific contribution to decadal surface temperature trends in the Arctic

Noel Keenlyside (UiB, BCCR, BCPU) Impacts of CGCM bias reduction on the equatorial Atlantic inter-annual variability

Predictability limits

Ingo Bethke (UiB, BCPU, Blue-Action) Improving statistical methods for assessing climate prediction skill

Torben Schmith (DMI, Blue-Action, EUCP) Semi-empirical improvement of seasonal forecasts of European winter temperatures,

Fei Li (NILU, BCPU) Subseasonal-to-Seasonal Forecasts with the Norwegian Climate Prediction Model

Bo Christiansen (DMI, Blue-Action, EUCP) The skill of dynamical decadal forecasts with focus on the North Atlantic region,

Stefan Sobolowski (NORCE) Investigating drivers of midlatitude circulation biases in climate reanalysis ensembles,

Leilane Passos (UiB, BCPU) Interannual to Decadal Predictions of Thermohaline Anomalies and Air-Sea Interaction in the Subpolar North Atlantic and the Nordic Seas,

Data assimilation

Sebastien Barthélemy (UiB, BCPU) Toward background error covariance hybridization for climate prediction,

Ali Aydogdu (NERSC) Data assimilation using adaptive, non-conservative, moving mesh models

Avneet Singh (BCPU) Optimising cross-covariance update in strongly coupled data assimilation

Francine Schevenhoven (BCPU) Efficient algorithms to train supermodels,

Julien Brajard (NERSC) Data assimilation as a machine learning tool or in combination with it to emulate a dynamical model from sparse and noisy observations,

Tian Tian (DMI, Blue-Action) The role of Arctic sea ice initialisation in decadal climate prediction: linking the Arctic sea ice loss and the mid-latitude climate,



Back to top