The Impact of Enhanced Vertical Resolution on Seasonal Prediction Skill of the Quasi‐Biennial Oscillation and Its Teleconnections in CESM2: A New Community Hindcast Dataset

July 7, 2026

A new paper entitled “The Impact of Enhanced Vertical Resolution on Seasonal Prediction Skill of the Quasi‐Biennial Oscillation and Its Teleconnections in CESM2: A New Community Hindcast Dataset” was recently published in the Journal of Geophysical Research: Atmospheres. This work represents a collaboration between the Center for Western Weather and Water Extremes (CW3E) at the Scripps Institution of Oceanography, the National Center for Atmospheric Research (NCAR), and the National Institute of Water and Atmospheric Research (NIWA), New Zealand. The paper is led by Isla Simpson (NCAR), co-authored by CW3E researchers Michael DeFlorio, Jiabao Wang, Nora Mascioli, Matthew Simpson, Patrick Mulrooney, Luca Delle Monache, and NCAR researchers Jim Edwards, Nan Rosenbloom, Gokhan Danabasoglu, Yuanpu Li, Stephen Yeager, Jadwiga Richter, and NIWA researcher Peter Gibson. This research, supported by the California Department of Water Resources and U.S. Army Engineer Research and Development Center, introduces a new seasonal hindcast dataset with the Community Earth System Model version 2 (CESM2) with a raised atmospheric model lid and enhanced vertical resolution. The dataset is freely available to the research community and provides improved representation of the stratosphere for seasonal prediction studies. This work supports the Advanced Precipitation and Streamflow Prediction priority in CW3E’s 2025-2029 Strategic Plan by advancing understanding of “forecasts of opportunity” where sub-seasonal to seasonal prediction skill can be improved.

This study compares the seasonal prediction skill of two newly developed initialized CESM2.1 hindcast datasets. One uses the enhanced vertical resolution of CESM3 in the troposphere and stratosphere (referred to as L83), while the other uses the vertical resolution of the previous-generation CESM2 configuration (referred to as L32). The comparison examines how increased vertical resolution and a raised model top influence seasonal prediction of the Quasi-Biennial Oscillation (QBO) and climate features linked to the QBO, including the stratospheric polar vortex, North Atlantic Oscillation (NAO), Madden-Julian Oscillation (MJO), North Pacific subtropical jet, Tropical Easterly Jet, and hydroclimate variables such as precipitation and temperature.

The enhanced vertical resolution and raised model lid substantially improve prediction of the QBO. In particular, the L83 configuration better captures the evolution of the QBO, including the downward propagation of lower-stratospheric zonal wind anomalies and the subsequent descent of the opposite-sign phase from above with amplitudes comparable to those observed in ERA5 (Figure 1). However, the improved QBO prediction skill in L83 does not translate into notable improvements in predicting the winter stratospheric polar vortex through the Holton-Tan effect. The results suggest a potential signal-to-noise problem in the model, whereby the Holton-Tan relationship is too weak and/or the stratospheric polar vortex exhibits excessive variability unrelated to the QBO. The lack of improvement in polar vortex prediction also results in no significant improvement in NAO prediction skill in L83 relative to L32.

The MJO is a convectively coupled disturbance that propagates eastward from the Indian Ocean toward the eastern Pacific on intraseasonal timescales. Through atmospheric teleconnections, MJO can affect weather patterns in the extratropics. Improved prediction of MJO characteristics on seasonal timescales could therefore enhance prediction skill worldwide. Observations indicate that the QBO strongly modulates MJO activity, with stronger MJO occurring during the easterly phase of the QBO. However, neither hindcast dataset reproduces this observed relationship (Figure 2). Because the L83 configuration demonstrates very high skill in representing the QBO itself, these results suggest that accurately simulating the QBO alone is insufficient to reproduce its influence on the MJO. Likewise, the improved fidelity of the QBO in L83 does not lead to enhanced prediction skill for upper-tropospheric Pacific zonal winds, tropical precipitation, or the Tropical Easterly Jet.

These seasonal hindcast datasets can be used for further investigation into seasonal prediction skill with CESM and the potential role of the stratosphere. Evidence shown in this study suggests a broader signal-to-noise issue in the model’s representation of QBO teleconnections, whereby observed relationships between the QBO and various climate features are stronger than those simulated by the model. Improving the representation of these teleconnections could unlock additional seasonal prediction skill for these climate features and their associated surface impacts.

Figure 1. Composites of daily averaged 5° S to 5° N [U] based on the ERA5 phase of the QBO at 50 hPa on the 1st November. Westerly QBO (QBOW) hindcasts are averaged to produce (a)–(c) and these are defined as those when the ERA5 5° S to 5° N [U] at 50 hPa is greater than +1σ where σ is the interannual standard deviation from 1970 to 2020 of the 50 hPa 5° S to 5° N [U] on Nov 1st. (d)–(f) show the average of Easterly QBO (QBOE) hindcasts, which are defined as those where the ERA5 5° S to 5° N [U] at 50 hPa on 1st Nov is less than – 1σ. Figure 3 from Simpson et al. 2026.

Figure 2. (a)–(c) the climatological DJF standard deviation of daily MJO‐filtered OLR (σ(OLRMJO)) for (a) ERA5, (b) L83, and (c) L32. (d)–(f) are as (a)–(c) but for the difference in σ(OLRMJO) between QBOE and QBOW DJF seasons. All panels use 1979 to 2020. Adapted from Figure 13 Simpson et al. 2026.

Citation:

Simpson, I. R., DeFlorio, M. J., Wang, J., Mascioli, N. R., Edwards, J., Rosenbloom, N., Simpson, M., Danabasoglu, G., Li, Y., Gibson, P. B., Mulrooney, P., Yeager, S. G., Richter, J. H., & Delle Monache, L. (2026). The Impact of Enhanced Vertical Resolution on Seasonal Prediction Skill of the Quasi‐Biennial Oscillation and Its Teleconnections in CESM2: A New Community Hindcast Dataset. Journal of Geophysical Research: Atmospheres, 131(12). https://doi.org/10.1029/2025jd046148