Training a supermodel with noisy and sparse observations: a case study with CPT and the synch rule on SPEEDO – v.1

Schevenhoven, F., Carrassi, A. 2022: Training a supermodel with noisy and sparse observations: a case study with CPT and the synch rule on SPEEDO – v.1. Geosci. Model Dev. https://doi.org/10.5194/gmd-15-3831-2022

Summary: As an alternative to using the standard multi-model ensemble (MME) approach to combine the output of different models to improve prediction skill, models can also be combined dynamically to form a so-called supermodel. The supermodel approach enables a quicker correction of the model errors. In this study we connect different versions of SPEEDO, a global atmosphere-ocean-land model of intermediate complexity, into a supermodel. We focus on a weighted supermodel, in which the supermodel state is a weighted superposition of different imperfect model states. The estimation, “the training”, of the optimal weights of this combination is a critical aspect in the construction of a supermodel. In our previous works two algorithms were developed: (i) cross pollination in time (CPT)-based technique and (ii) a synchronization-based learning rule (synch rule). Those algorithms have so far been applied under the assumption of complete and noise-free observations. Here we go beyond and consider the more realistic case of noisy data that do not cover the full system’s state and are not taken at each model’s computational time step. We revise the training methods to cope with this observational scenario, while still being able to estimate accurate weights. In the synch rule an additional term is introduced to maintain physical balances, while in CPT nudging terms are added to let the models stay closer to the observations during training. Furthermore, we propose a novel formulation of the CPT method allowing the weights to be negative. This makes it possible for CPT to deal with cases in which the individual model biases have the same sign, a situation that hampers constructing a skillfully weighted supermodel based on positive weights. With these developments, both CPT and the synch rule have been made suitable to train a supermodel consisting of state of the art weather and climate models.

Link to publication. You are most welcome to contact us or the corresponding author(s) directly, if you have questions.

On the trail of the disappearing polar sea ice

Jakob Dörr is a PhD student at the University of Bergen, working with Marius Årthun in the BCPU research area on “Understanding mechanisms for climate predictability”. In the spring of 2022, he travelled to California, to visit Dave Bonan. For four weeks, Jakob had the opportunity to work with Dave and his working group at the California Institute of Technology (Caltech) in Pasadena. Dave is also a PhD student, and also interested in understanding present and future changes in the Earth’s sea ice cover, and which processes these are caused by.

In the Arctic, the sea ice cover has strongly declined in all seasons over the last 40 years, and this is mostly due to the Earth’s warming, driven by anthropogenic greenhouse gas emissions. However, because the sea ice interacts with the ocean and the atmosphere and is thus part of the chaotic climate system, it is affected by random fluctuations and internal variability which is independent from the long-term warming trend. These variability modes can affect the sea ice for periods of up to several decades. It is therefore not entirely clear exactly how much of the sea ice loss we have observed over the last decades was due to global warming, and how much was because of internal variability. Dave and Jakob are working to detect and separate those variability modes that affect the sea ice over long periods (decades and longer) in the observational record of sea ice. They use a novel technique developed by Robb Wills at the University of Washington.

Jakob and Dave are hoping to determine for different regions of the Arctic, which modes of variability affect the sea ice cover, and how much their influence compares to the long-term sea ice loss due to global warming. This will help to understand and attribute past sea ice changes and enhance our ability to predict the future regional sea ice loss. While Jakob focuses on the Arctic, Dave applies the same technique to the Antarctic, where a steady increase in sea ice cover over the last decades, followed by a strong decline since 2016, has been observed. Their analysis might shed some light on the mechanisms behind this puzzling evolution, and how much of it was caused by changes due to global warming. The goal of the BCPU-supported visit was to prepare work for two separate publications on Arctic and Antarctic sea ice, respectively, and to discuss how experience from observations can be applied to analyse climate model simulations of future sea ice change.


“During the visit, we exchanged our experience and discussed new ideas for our analysis. I also got to meet scientists in both the Oceanography (Andrew Thompson) and the Climate Dynamics (Tapio Schneider) group at Caltech. I was also lucky to come at a time where Caltech was opening fully again, with many international scientists visiting the institute. Furthermore, I got invited to be part of a sea ice reading course where we had intense discussions about sea ice models, trends and mechanisms with Dave and other members of the Oceanography group. On top of that, I had the chance to visit some friends from the Scripps Institute of Oceanography in San Diego. I had a lot of interactions during the visit and learned a lot about how science is conducted at Caltech and other US institutions. I hope that I can continue the collaboration between Caltech and the Bjerknes Center beyond our work on sea ice observations. The visit showed me how important it is to physically meet people to exchange ideas and develop collaborations across the globe. There are plans that Dave visits us in Bergen next spring, and I hope to return to Pasadena after that.” – Jakob Dörr

Predicting Arctic sea ice for the marine transport sector

Image credit: Copernicus Marine Service

Check out this nice article by Dr. Ellen Viste at the Bjerknes Centre for Climate Research, on the evaluation of sea ice models and how far we are to being able to provide reliable, near-term sea-ice predictions:

Read the article: Predicting Arctic sea ice.

In it, we hear from Tarkan Bilge, our BCPU data manager, and his recent paper on sea ice thickness forecasts to support Arctic marine transport, together with other collaborating scientists at our partner, NERSC, among others.

WMO Global Annual to Decadal Climate Update: A Prediction for 2021–25

Hermanson, L., Smith, D., Seabrook, M., Bilbao, R., Doblas-Reyes, F., Tourigny, E., Lapin, V., Kharin, V.V., Merryfield, W.J., Sospedra-Alfonso, R., Athanasiadis, P., Nicoli, D., Gualdi, S., Dunstone, N., Eade, R., Scaife, A., Collier, M., O’Kane, T., Kitsios, V., Sandery, P., Pankatz, K., Früh, B., Pohlmann, H., Müller, W., Kataoka, T., Tatebe, H., Ishii M., Imada, Y., Kruschke, T., Koenigk, T., Pasha Karami, M., Yang, S., Tian, T., Zhang, L., Delworth, T., Yang, X., Zeng, F., Wang, Y., Counillon, F., Keenlyside, N.S., Bethke, I., Lean, J., Luterbacher, J., Kumar Kolli, R., Kumar, A. 2022: WMO Global Annual to Decadal Climate Update: A Prediction for 2021–25. BAMS https://doi.org/10.1175/BAMS-D-20-0311.1 .

Summary: As climate change accelerates, societies and climate-sensitive socioeconomic sectors cannot continue to rely on the past as a guide to possible future climate hazards. Operational decadal predictions offer the potential to inform current adaptation and increase resilience by filling the important gap between seasonal forecasts and climate projections. The World Meteorological Organization (WMO) has recognized this and in 2017 established the WMO Lead Centre for Annual to Decadal Climate Predictions (shortened to “Lead Centre” below), which annually provides a large multimodel ensemble of predictions covering the next 5 years. This international collaboration produces a prediction that is more skillful and useful than any single center can achieve. One of the main outputs of the Lead Centre is the Global Annual to Decadal Climate Update (GADCU), a consensus forecast based on these predictions. This update includes maps showing key variables, discussion on forecast skill, and predictions of climate indices such as the global mean near-surface temperature and Atlantic multidecadal variability. it also estimates the probability of the global mean temperature exceeding 1.5°C above preindustrial levels for at least 1 year in the next 5 years, which helps policy-makers understand how closely the world is approaching this goal of the Paris Agreement. This paper, written by the authors of the GADCU, introduces the GADCU, presents its key outputs, and briefly discusses its role in providing vital climate information for society now and in the future..

Link to publication. You are most welcome to contact us or the corresponding author(s) directly, if you have questions.

Propagation of Thermohaline Anomalies and Their Predictive Potential along the Atlantic Water Pathway

Langehaug, H. R., Ortega, P., Counillon, F., Matei, D., Maroon, E., Keenlyside, N., Mignot, J., Wang, Y., Swingedouw, D., Bethke, I., Yang, S., Danabasoglu, G., Bellucci, A., Ruggieri, P., Nicolì, D., Årthun, M. 2022: Propagation of Thermohaline Anomalies and Their Predictive Potential along the Atlantic Water Pathway. J Clim. https://doi.org/10.1007/s10236-022-01523-x

Summary: In this study, we find that dynamical prediction systems and their respective climate models struggle to realistically represent ocean surface temperature variability in the eastern subpolar North Atlantic and Nordic seas on interannual-to-decadal time scales. In previous studies, ocean advection is proposed as a key mechanism in propagating temperature anomalies along the Atlantic water pathway toward the Arctic Ocean. Our analysis suggests that the predicted temperature anomalies are not properly circulated to the north; this is a result of model errors that seems to be exacerbated by the effect of initialization shocks and forecast drift. Better climate predictions in the study region will thus require improving the initialization step, as well as enhancing process representation in the climate models.

Link to publication. You are most welcome to contact us or the corresponding author(s) directly, if you have questions.

Spatial patterns, mechanisms and predictability of Barents Sea ice change

Efstathiou, E., Eldevik, T., Årthun, M., Lind, S. 2022: Spatial patterns, mechanisms and predictability of Barents Sea ice change. J Clim. https://doi.org/10.1175/JCLI-D-21-0044.1  .

Summary: Recent Arctic winter sea ice loss has been most pronounced in the Barents Sea. Here we explore the spatial structure of Barents Sea ice change as observed over the last 40 years. The dominant mode of winter sea ice concentration interannual variability corresponds to areal change (explains 43% of spatial variance) and has a center of action in the northeastern Barents Sea where the temperate Atlantic inflow meets the wintertime sea-ice. Sea ice area import and northerly wind also contribute to this “areal-change mode”; the area increases with more ice import and stronger winds from the north. The remaining 57% variance in sea ice, individually and combined, redistributes the sea ice without changing the total area. The two leading redistribution modes are a dipole of increase in sea ice concentration south of Svalbard with decrease southwest of Novaya Zemlya, and a tripole of increase in the central Barents Sea with decrease east of Svalbard and in the southeastern Barents Sea. Redistribution is mainly contributed by anomalous wind and sea ice area import. Basic predictability, i.e., the lagged response to observed drivers, is predominantly associated with the areal-change mode as influenced by temperature of the Atlantic inflow and sea ice import from the Arctic.

Link to publication. You are most welcome to contact us or the corresponding author(s) directly, if you have questions.

Estimation of Ocean Biogeochemical Parameters in an Earth System Model Using the Dual One Step Ahead Smoother: A Twin Experiment

Singh, T., Counillon, F., Tjiputra, J., Wang Y., El Gharamti, M. 2022: Estimation of Ocean Biogeochemical Parameters in an Earth System Model Using the Dual One Step Ahead Smoother: A Twin Experiment. Front. Mar. Sci. https://doi.org/10.3389/fmars.2022.775394 .

For an easy-to-understand overview of this publication, produced in collaboration with the TRIATLAS project, we recommend starting with this neat article written by Henrike Wilborn, at NERSC: “Making climate models more accurate by improving their tuning“.

Summary: Ocean biogeochemical (BGC) models utilise a large number of poorly-constrained global parameters to mimic unresolved processes and reproduce the observed complex spatio-temporal patterns. Large model errors stem primarily from inaccuracies in these parameters whose optimal values can vary both in space and time. This study aims to demonstrate the ability of ensemble data assimilation (DA) methods to provide high-quality and improved BGC parameters within an Earth system model in an idealized perfect twin experiment framework. We use the Norwegian Climate Prediction Model (NorCPM), which combines the Norwegian Earth System Model with the Dual-One-Step ahead smoothing-based Ensemble Kalman Filter (DOSA-EnKF). We aim to estimate five spatially varying BGC parameters by assimilating salinity and temperature profiles and surface BGC (Phytoplankton, Nitrate, Phosphate, Silicate, and Oxygen) observations in a strongly coupled DA framework—i.e., jointly updating ocean and BGC state-parameters during the assimilation. We show how BGC observations can effectively constrain error in the ocean physics and vice versa. The method converges quickly (less than a year) and largely reduces the errors in the BGC parameters. Some parameter error remains, but the resulting state variable error using the estimated parameters for a free ensemble run and for a reanalysis performs nearly as well as with true parameter values. Optimal parameter values can also be recovered by assimilating climatological BGC observations or sparse observational networks. The findings of this study demonstrate the applicability of the DA approach for tuning the system in a real framework.

Link to publication. You are most welcome to contact us or the corresponding author(s) directly, if you have questions.

Dr. Helene Langehaug explains our work to the Norwegian Minister for Climate and the Environment

Photo by Gudrun Sylte

On January 31st 2022, the Bjerknes Centre for Climate Research received Espen Barth Eide, our minister for Climate and the Environment, for a visit. On this occasion, Dr. Helene Langehaug gave a talk about the role of the ocean in improving climate predictions, which is a crucial aspect the Bjerknes Climate Prediction Unit’s research work, towards improving predictions from seasons to several years ahead.
Find out more about our research areas and results, or just get in touch with our team.

Assessing the influence of sea surface temperature and arctic sea ice cover on the uncertainty in the boreal winter future climate projections

Cheung, HN., Keenlyside, N., Koenigk, T., Yang, S., Tian, T., Xu, Z., Gao, Y., Ogawa, F., Omrani, N.-E., Qiao, S., Zhou, W. 2022: Assessing the influence of sea surface temperature and arctic sea ice cover on the uncertainty in the boreal winter future climate projections. Clim. Dyn. https://doi.org/10.1007/s00382-022-06136-0

Summary: We investigate the uncertainty (i.e., inter-model spread) in future projections of the boreal winter climate, based on the forced response of ten models from the CMIP5 following the RCP8.5 scenario. The uncertainty in the forced response of sea level pressure (SLP) is large in the North Pacific, the North Atlantic, and the Arctic. A major part of these uncertainties (31%) is marked by a pattern with a center in the northeastern Pacific and a dipole over the northeastern Atlantic that we label as the Pacific–Atlantic SLP uncertainty pattern (PA∆SLP). The PA∆SLP is associated with distinct global sea surface temperature (SST) and Arctic sea ice cover (SIC) perturbation patterns. To better understand the nature of the PA∆SLP, these SST and SIC perturbation patterns are prescribed in experiments with two atmospheric models (AGCMs): CAM4 and IFS. The AGCM responses suggest that the SST uncertainty contributes to the North Pacific SLP uncertainty in CMIP5 models, through tropical–midlatitude interactions and a forced Rossby wavetrain. The North Atlantic SLP uncertainty in CMIP5 models is better explained by the combined effect of SST and SIC uncertainties, partly related to a Rossby wavetrain from the Pacific and air-sea interaction over the North Atlantic. Major discrepancies between the CMIP5 and AGCM forced responses over northern high-latitudes and continental regions are indicative of uncertainties arising from the AGCMs. We analyze the possible dynamic mechanisms of these responses, and discuss the limitations of this work.

Link to publication. You are most welcome to contact us or the corresponding author(s) directly, if you have questions.

Changes in Arctic Stratification and Mixed Layer Depth Cycle: A Modeling Analysis

Hordoir, R., Skagseth, Ø., Ingvaldsen, R.B., Sandø, A.B., Löptien, U., Dietze, H., Gierisch, A.M.U., Assmann K.A., Lundesgaard,Ø., Lind, S. 2022: Changes in Arctic Stratification and Mixed Layer Depth Cycle: A Modeling Analysis. JGR Oceans. https://doi.org/10.1029/2021JC017270

Summary: We analyzed the results of an ocean model simulation for the Arctic and North Atlantic oceans for the period 1970–2019. Our model is in line with the recent observed changes in the Arctic Ocean and allows, in contrast to the rather sparse observations, a detailed assessment of stratification changes. These changes will affect the Arctic ecosystem and are also believed to affect the large scale ocean circulation. We show that major changes in upper ocean conditions are caused by changes in the fresh water supply by sea ice and varying effect of the wind on regions that are now becoming ice-free. We also study the effect of changes in river runoff into the Arctic Ocean. Our study shows that an increase in river runoff can change the coastal circulation and results, paradoxically, in regions of higher salinity. These results point to the importance of modeling tools when it comes to a better understanding of ocean processes in a changing climate.

Link to publication. You are most welcome to contact us or the corresponding author(s) directly, if you have questions.