<scp>QBOi</scp> ‐ <scp>SNAP</scp> ‐ <scp>QUOCA</scp> workshop: improved simulations of the stratosphere for better predictions of weather, climate and extreme events

In the week of 24 March 2025, the first joint workshop between the Quasi-Biennial Oscillation initiative (QBOi), Stratospheric Network for the Assessment of Predictability (SNAP) and QUasi-biennial oscillation and Ozone Chemistry interactions in the Atmosphere (QUOCA) was held at the Isaac Newton Institute for Mathematical Sciences in Cambridge. QBOi, SNAP and QUOCA are all part of APARC (Atmospheric Processes and their Role in Climate), a core project of the World Climate Research Programme. All three activities focus on stratospheric dynamics: QBOi on improving climate model representation of the QBO and its teleconnections; SNAP on the ability of subseasonal-to-seasonal (S2S) forecast systems to simulate the stratosphere and its tropospheric coupling; QUOCA on ozone feedbacks on the QBO. The workshop aimed to improve understanding of stratospheric processes, variability, uncertainties and influence on surface climate and predictability. There was a diversity of scientists participating in the workshop (Figure 1): there were 105 people from 74 institutes across 18 countries, with women comprising around a third and early careers over half of all attendees, and almost 80% attending in-person. There were 54 posters and 55 talks on 4 key topics described below. The programme also included QBOi and SNAP activity reporting, the in-person QUOCA kick-off, breakouts and plenary sessions to facilitate community input into activity planning and potential opportunities for cross-activity collaborations, and various social activities (Figure 2). This session highlighted challenges in representing stratosphere–troposphere interactions in climate models, focusing on connections between the QBO, Madden-Julian Oscillation (MJO) and El Niño-Southern Oscillation (ENSO). Several talks addressed the difficulty of modelling the influence of the QBO and extratropical forcings, such as sudden stratospheric warmings (SSWs), on the tropical stratosphere (Shunsuke Noguchi). A large proportion of talks covered the QBO interaction with the MJO, raising questions about model biases and the physical understanding of QBO–MJO coupling (Lon Hood, Kai Huang, Chang-Hyun Park). Others showed how correcting ozone profiles in climate models led to a reduction in temperature and moisture biases; however, the QBO–MJO connection seemed to be particularly sensitive to models’ convective parametrisations (Seok-Woo Son, Jiyoung Oh, Seung-Yoon Back). The potential sensitivity of the MJO to ENSO was also highlighted (Raina Roy, Yoshio Kawatani, Dillon Elsbury). This session explored tropospheric coupling mechanisms (Wuhan Ning, Mark Baldwin, Yueyue Yu). A number of talks discussed the range of abilities of models in representing some aspects of downward coupling, both in S2S and climate models (Jian Rao, Jiankai Zhang, Rongzhao Lu, Simon Lee, Alexey Karpechko). The Stratospheric Nudging and Predictable Surface Impacts (SNAPSI) project emerged as a focus of this session. It aims to assess the added forecast skill from SSWs in S2S models. Most models predicted downward impacts following the SSW case studies but varied in their magnitude/timing, with other discussed deficiencies (Hera Kim, Peter Hitchcock, Jinlong Huang). While the 2018 and 2019 SSWs both drove negative NAO regimes, their downward propagation differed (Robert Lee, Dong-Chan Hong). The 2018 SSW was also found to increase the risk of Iberian extreme rains, European cold spells, excess deaths and high energy demand (William Seviour, Ying Dai, Regan Mudhar). A new method for identifying atmospheric weather regimes that are particularly useful for impacts was also described (Marlene Kretschmer). Overall, this session emphasised how improved stratospheric modelling can enhance weather and climate predictability – though biases remain. Machine learning (ML) was suggested to help, but is limited by sparse stratospheric training data (Inna Polichtchouk). This session explored how stratospheric dynamics and composition influence each other. For example, ozone-climate feedbacks emerged as a crucial driver of stratospheric temperature trends (Gabriel Chiodo), while ozone distributions were found to be influenced by Brewer-Dobson circulation (BDC) responses to different QBO phases (Veenus Venugopal, Alison Ming). A conceptual model to unify various processes was also introduced, linking surface warming and QBO phase shifts to ozone distribution variations (Aaron Match). A number of talks discussed the need for accurate representation of stratospheric polar vortex (SPV)-related mechanisms for predictability and variability. This included the stratospheric response to Arctic sea-ice loss (Xiaocen Shen), ozone-circulation interactions (Siyi Zhao), and the influence of wildfire and volcanic eruption-driven ozone anomalies (Eun-Pa Lim). These overall highlighted the need for improved QBO and ozone chemistry representation for climate model projections and S2S predictability (Zhe Wang). Some research utilised observation-based products, with speakers highlighting their associated uncertainties as well as potential areas for model evaluation and improved gravity wave (GW) parametrisations (Tobias Kerzenmacher, Kimberlee Dubé, Aleena Moolakkunnel Jaison). Another interesting theme was geoengineering: the impacts of stratospheric aerosol injection experiments on lower-stratospheric heating and large-scale circulation changes seemed to underscore the importance of interactive ozone chemistry as well as the injection location (Ewa Bednarz, Colleen Golja). There was significant interest in the news that monitoring of the atmosphere with high-vertical resolution sounders may cease by 2027. Observations are necessary for modelling; QBO representation in models remains a challenge and depends on GW parametrisations. A number of talks introduced alternative methods, such as new data processing methods (Corwin Wright) and model parametrisations (Martina Bramberger, Pu Lin). Others evaluated the performance of new versions of climate models in simulating the QBO (Mijeong Park, Lawrence Coy), including discussion of the importance of considering vertical resolution both for the QBO and polar vortex biases (Natasha Trencham, Wandi Yu). A couple of talks touched on the use of ML to support model tuning, though it was noted that significant training data are required (Aditi Sheshadri, Walter Hannah). Several talks focused on teleconnections; there was found to be a strong dependence of these on how models simulate the QBO, including its amplitude and location, and, for northern hemisphere winter especially, its interaction with ENSO (Martin Andrews, Vinay Kumar, Froila Palmeiro). Some cross-activity talks discussed the ability of SNAPSI models to represent the QBO (Hamid Pahlavan, Yue Wang) – including their inability to simulate QBO disruptions, which had elsewhere been simulated in Large Ensemble Single Forcing MIP experiments following large volcanic forcing events (Chaim Garfinkel). The sensitivity of the QBO to CO2 and ozone increases was also shown in single-model analyses (Hyun-Kyu Lee, Christiane Jablonowski). This joint meeting successfully enabled the communication of the most up-to-date science across the activities. Several science highlights arose over the week: observations, ozone, GW parametrisations, SNAPSI, stratosphere–troposphere coupling and other teleconnections, and ML. Attendees also found the breakouts and plenary discussions particularly valuable, with many enjoying this explicit time for discussion. In the breakouts, pre-planned prompts generated engaging discussions on the current and future role of stratospheric research in climate science, identifying gaps in our current knowledge to inform further coordinated model experiments (Figure 3). There was significant interest in the potential of ML to enhance the representation of stratospheric dynamics, large-scale circulation and stratosphere–troposphere coupling. Attendees were generally optimistic about the evolution and use of ML in parametrisation and improving model skill. However, there was a shared concern that increased reliance on data-driven methods could come at the cost of physical interpretability; there was support for a hybrid modelling approach that combines ML with physics-based frameworks. Across most breakout discussions, the need for expanded observational datasets was evident; the current record has limited temporal and spatial coverage, but observations are necessary for constraining models and evaluating ML approaches. Elsewhere, some conversations highlighted shortcomings in model simulations, including increased resolution and interactive chemistry. There was also a general agreement to improve accessibility and awareness of available data; a post-workshop survey revealed that the majority of attendees (~60%) utilised JASMIN (the UK’s collaborative data analysis environment; Lawrence et al., 2013) for downloading, processing and/or analysing data. Each of the QSQ activities is organised into multiple working groups to address key questions around model representation of the stratosphere and broader implications for biases and predictive skill. Many of these groups are actively developing research papers to synthesise progress and identify future directions. As Peter Haynes and Martin Chipperfield summarised on the final day, the workshop demonstrated the community’s continued focus on S2S prediction, better simulations of the QBO, the importance of (ozone) chemistry and its interactions, as well as links between meteorological events and stratospheric conditions. Attendees tended to support continued focus on improving the ability to simulate and predict stratospheric variability, which remains essential for advancing model performance, interpreting observational trends and understanding the long-term behaviour of the climate. Through these activities and projects, this community is laying the groundwork for more accurate and comprehensive representations of the stratosphere in climate models, paving the way for improved prediction and understanding of the climate system. You can learn more about and find contact information for each of the activities at https://www.aparc-climate.org/activities/. Regan Mudhar: writing – original draft preparation; writing – review and editing. Anna Hall: original draft preparation; writing – review and editing; visualization. Jinlong Huang: original draft preparation; writing – review and editing. Robert W. Lee: original draft preparation; writing – review and editing. Froila M. Palmeiro: original draft preparation; writing – review and editing. We would like to thank the workshop organising committee: Alison Ming (University of Cambridge, UK), Neal Butchart (Met Office Hadley Centre, UK), Scott Osprey (University of Oxford, UK), James Anstey (Canadian Centre for Climate Modelling and Analysis, Canada), Chaim Garfinkel (The Institute of Earth Sciences, Israel), Clara Orbe (NASA Goddard Institute for Space Studies, USA), Amy Butler (NOAA Chemical Sciences Laboratory, USA), Peter Hitchcock (Cornell University, USA) and Yoshio Kawatani (Hokkaido University, Japan). We also gratefully acknowledge the University of Cambridge, the Isaac Newton Institute for Mathematical Sciences and the Institute of Computing for Climate Science for supporting the workshop. We acknowledge the contribution to workshop funding from the Met Office Hadley Centre Climate Programme funded by DSIT. We also acknowledge funding for participant travel support from the US National Science Foundation under grant number 2516533. Financial support was also received from the International Union of Geodesy and Geophysics (IUGG) and APARC. R. Mudhar is Co-Editor-in-Chief of Weather but was not privy to the review process.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.