Wu et al. (2026) Decomposition of Pacific Decadal Oscillation using linear inverse models sheds light on its dominant modes and future response

Identification

• Journal: npj Climate and Atmospheric Science
• Year: 2026
• Date: 2026-01-10
• Authors: Sheng Wu, Di Lorenzo Emanuele, Yingying Zhao, Matthew Newman, Zhengyu Liu, Antonietta Capotondi, Daoxun Sun, Samantha Stevenson, Yonggang Liu
• DOI: 10.1038/s41612-025-01315-2

Research Groups

• Department of Earth, Environmental, and Planetary Sciences, Brown University, Providence, RI, USA
• Laboratory for Climate and Ocean-Atmosphere Studies & Department of Atmospheric and Oceanic Sciences, School of Physics, Peking University, Beijing, China
• Laoshan Laboratory, Qingdao, China
• Physical Sciences Laboratory, NOAA, Boulder, CO, USA
• Atmospheric Science Program, Department of Geography, Ohio State University, Columbus, OH, USA
• Bren School of Environmental Science and Management, University of California, Santa Barbara, CA, USA
• Institute of Ocean Research, Peking University, Beijing, China

Short Summary

This study decomposes the Pacific Decadal Oscillation (PDO) into three dynamical modes using a Linear Inverse Model (LIM) to understand its mechanisms and project its future response under climate change. It finds that under global warming, the El Niño–Southern Oscillation (ENSO) mode's contribution to PDO increases while the Kuroshio-Oyashio Extension (KOE) mode's influence diminishes, leading to a shortened PDO timescale.

Objective

• To decompose the Pacific Decadal Oscillation (PDO) into its dominant dynamical constituents (Kuroshio-Oyashio Extension (KOE) mode, North Pacific–Central Tropical Pacific (NP-CP) mode, and El Niño–Southern Oscillation (ENSO) mode) using a Linear Inverse Model (LIM).
• To assess the relative importance of these modes in observations and climate models, identifying systematic model biases in representing their spatial structures.
• To project the future response of PDO and its constituent modes under global warming scenarios (SSP585) and elucidate the underlying mechanisms for changes in PDO timescale.
• To determine the minimum data length required for robust analysis of PDO's decadal variability.

Study Configuration

• Spatial Scale: Global, North Pacific (20°N-60°N, 120°E-120°W), Tropical Pacific (20°S-20°N, 100°E-100°W), Kuroshio-Oyashio Extension (KOE) region.
• Temporal Scale: Decadal and interannual variability; observational data (HadISST) from 1920-2014 (85 years); climate model historical runs (1920-2005, 85 years); future projections (2015-2100, 85 years); LIM large ensemble integrations extended to 200, 300, 400, 500, and 1000 years for robustness analysis. Timescales of identified modes: KOE (multi-decadal/infinite), NP-CP (decadal, 30-50 years), ENSO (interannual, 2-8 years).

Methodology and Data

• Models used:
  • Linear Inverse Model (LIM) for decomposition, reconstruction, and large ensemble generation (LIM-LE).
  • Community Earth System Model Large Ensemble (CESM-LE) for "same physical processes and anthropogenic forcing" system (SSA).
  • Coupled Model Intercomparison Project Phase 6 (CMIP6) multi-model ensemble for "different physical processes and anthropogenic forcing" system (SDA).
• Data sources:
  • Hadley Centre Sea Ice and Sea Surface Temperature (HadISST) dataset (1870-2012, 1°×1° spatial resolution).
  • NOAA’s Extended Reconstructed Sea Surface Temperature dataset version 4 (ERSST.v4) for robustness assessment.
  • CESM-LE output (25 members, 1920-2005 historical run).
  • CMIP6 historical run (CMIP6 HIS) and Shared Socioeconomic Pathway 5-8.5 (CMIP6 SSP585) scenario output (25 models).

Main Results

• The Pacific Decadal Oscillation (PDO) is dynamically decomposed into three main constituents: the Kuroshio-Oyashio Extension (KOE) mode, the North Pacific–Central Tropical Pacific (NP-CP) mode, and the El Niño–Southern Oscillation (ENSO) mode.
• In observations (1920-2014), the relative importance of these modes to PDO is 0.31 for KOE, 0.53 for NP-CP, and 0.16 for ENSO.
• Climate models (CESM-LE and CMIP6 HIS) exhibit systematic biases in representing the spatial structures of the KOE and NP-CP modes, showing weaker tropical-extratropical teleconnections and stronger KOE local signals compared to observations.
• The relative importance of these modes varies substantially over 85-year periods due to internal climate variability; at least 300 years of data are required for statistically stable and stationary estimates of PDO dynamical constituents.
• Under global warming (CMIP6 SSP585 scenario), models project a significant increase in the relative importance of the ENSO mode (from 23-25% in historical to 35-41% in SSP585) and a decrease in the KOE mode's influence (from 38-42% to 26-28%). The NP-CP mode's contribution remains relatively unchanged.
• This shift in modal importance leads to a projected shortening of the PDO timescale under global warming, with spectral peaks shifting from decadal (25-30 years) towards higher frequencies (~12 years).

Contributions

• Provides a novel LIM-based dynamical decomposition of PDO into three distinct, physically interpretable modes (KOE, NP-CP, ENSO), facilitating more direct comparisons of PDO mechanisms between models and observations.
• Quantifies the significant uncertainty introduced by limited observational data length (85 years) for decadal variability analysis, establishing a lower bound of 300 years for robust mechanistic studies of PDO.
• Elucidates the mechanism for the projected shortening of the PDO period under global warming, attributing it to the increasing dominance of the ENSO mode and decreasing influence of the KOE mode.
• Highlights systematic biases in climate models' representation of PDO constituent spatial structures, offering a diagnostic tool for evaluating and improving climate model performance regarding decadal variability.

Funding

• National Natural Science Foundation of China (Grant Number 42225606)
• RGMA DOE Grant DE‐SC0023228
• National Natural Science Foundation of China (Grant Number 42405051)
• National Key R&D Program of China (Grant Number 2023YFF0805102)
• National Key R&D Program of China (Grant Number 2023YFF0805200)
• Taishan Scholars Program (No. tsqn202306298)
• Taishan Scholars Program (No. tsqn202306299)

Citation

@article{Wu2026Decomposition,
  author = {Wu, Sheng and Emanuele, Di Lorenzo and Zhao, Yingying and Newman, Matthew and Liu, Zhengyu and Capotondi, Antonietta and Sun, Daoxun and Stevenson, Samantha and Liu, Yonggang},
  title = {Decomposition of Pacific Decadal Oscillation using linear inverse models sheds light on its dominant modes and future response},
  journal = {npj Climate and Atmospheric Science},
  year = {2026},
  doi = {10.1038/s41612-025-01315-2},
  url = {https://doi.org/10.1038/s41612-025-01315-2}
}