Emerging trends in landfalling atmospheric rivers over the South Pacific

March 4, 2026

A new paper entitled “Emerging trends in landfalling atmospheric rivers over the South Pacific” was recently published in npj Climate and Atmospheric Science (2026). The study was led by Peter B. Gibson (Earth Sciences New Zealand), with contributions from co-authors Neelesh Rampal (Earth Sciences New Zealand); Felix W. Goddard (University of Canterbury, New Zealand); Bin Guan (UCLA/JPL); Michael J. DeFlorio (CW3E); and Hamish D. Prince (University of Wisconsin–Madison). Peter Gibson is an alumnus of the CW3E Subseasonal & Seasonal (S&S) team, and Hamish Prince participated in the CW3E Summer Intern Program in 2020.

This study investigates historical and projected changes in landfalling atmospheric rivers (ARs) across the South Pacific, with a focus on midlatitude land regions. Global climate models project that this region will be a hotspot for some of the largest increases in ARs on the planet, despite receiving comparatively less attention than Northern Hemisphere AR regions. [JK1.1]This research supports CW3E’s Strategic Plan (Atmospheric Rivers and Extreme Precipitation Research, Prediction, and Applications), by partnering with other academic institutions to advance research on ARs and extreme precipitation globally.

The analysis combines multiple AR detection methodologies applied to reanalysis datasets and high-resolution dynamical downscaling of climate model projections. The high-resolution downscaled simulations provide 12km resolution over New Zealand, and ~12-30km over the wider South Pacific region (Figure 1). This enhanced resolution over the wider region provides benefits for representing ARs across their lifecycle before making landfall.

Figure 1. Atmospheric resolution of the CCAM model used for downscaling. CCAM implements a stretched grid to focus high resolution over New Zealand (white shading) and the wider South Pacific. Figure 1 from Gibson et al. (2026).

A key finding from reanalysis-derived trends over the historical period from 1960 to 2023 shows that statistically significant increases in atmospheric river (AR) frequency are most robust over the Southern Ocean (approximately 45–60°S), where trends are spatially coherent and exceed internal variability. Over land, AR frequency trends are less spatially robust in reanalysis products; however, AR tracking approaches based on fixed percentiles of integrated vapor transport (IVT) reveal stronger increasing signals in parts of southern New Zealand and Tasmania. In contrast, AR tracking algorithms that emphasize synoptic-scale events suggest that AR frequency trends over land are not yet robustly detectable in historical reanalysis.

Another key finding from this work leverages high-resolution regional climate projections. These model simulations suggest that landfalling AR frequency and extreme AR occurrence are expected to become more widespread and statistically detectable within the next 10–20 years, with the earliest emergence projected over southern New Zealand, particularly during winter and spring. Across the model ensemble, five of six (83%) downscaled simulations indicate increasing near-term signals in landfalling AR frequency, with inter-model spread reflecting contributions from both thermodynamic processes (such as rising temperatures and increased water vapor) and dynamic processes (including jet stream changes). Under moderate emissions pathways, the frequency of extreme landfalling ARs (classified as AR5 events) could double by mid-century, implying heightened risks of AR-related hazards such as flooding and landslides in vulnerable regions.

A number of important research questions remain: Are trends in ARs from reanalysis already robustly detectable across this region, what role do thermodynamic and dynamic trends contribute? Can downscaled climate models capture these historical trends? When and where do models project that AR trends will become more widespread and robustly detectable across the 21st century? Future work from using this model ensemble will further investigate linkages between ex-tropical cyclones and ARs, as well explore trends in back-to-back ARs (also known as ‘AR families’) which can be particularly hazardous. Our results also indicate that further narrowing the uncertainty in AR projections will need to involve narrowing the uncertainty in how the jet stream responds to climate change across the Southern Hemisphere mid-latitudes.

Figure 2. Decade of emergence for extreme ARs. Decade of emergence for extreme ARs (rank AR5) showing the decade when there is first projected to be a doubling in frequency (panels a and c) and a 5x increase in frequency (panels b and d) under SSP2-4.5 and SSP3-7.0. The historical frequency of extreme ARs (rank AR5) is shown in panel e for refence, computed as the return period across all pooled simulations. Figure 14 from Gibson et al. (2026).

Citation:

Gibson, P. B., Rampal, N., Goddard, F. W., Guan, B., DeFlorio, M. J., & Prince, H. D. (2026). Emerging trends in landfalling atmospheric rivers over the South Pacific. npj Climate and Atmospheric Science (published online ahead of print 2026). https://doi.org/10.1038/s41612-026-01338-3