Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.23094

Graph-Learned Spring SST Evolution and Land-Ocean Synergy Improve Yangtze Summer Rainfall Prediction

Yifu Gao·Haishan Chen·Runze Li·Efi Foufoula-Georgiou·Jasper A. Vrugt
View original source ↗Read source document ↗Export metadata ↓

Abstract

Seasonal prediction of Yangtze summer rainfall is challenged by nonstationary oceanic precursors. Here we demonstrate that accounting for the evolution of spring SST and state-dependent land surface information improves predictability. Graph total variation (GTV) is applied to March-May SST fields over the western Pacific, Niño3.4, and tropical Indian Ocean to identify spatially coherent, month-specific SST coefficients for each grid cell and month to form a scalar GTV index. The learned coefficient patterns capture April-to-May ENSO decay and an opposing Indian Ocean adjustment, consistent with coupled Indo-Pacific variability. During 2008-2022, the GTV index prediction outperforms Niño3.4-based and sparse-regression benchmarks. Further improvement is achieved by including interactions of May Indo-China Peninsula soil moisture with Indian Ocean SST, importantly also correcting several underestimated wet summers. Circulation diagnostics associate the learned index with a summer geopotential-height dipole and Yangtze moisture convergence, while dry-Indochina/warm-Indian-Ocean conditions coincide with stronger westward extension of the western Pacific subtropical high.

Explore related subjects

physics.ao-ph

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Yifu Gao, Haishan Chen, Runze Li, Efi Foufoula-Georgiou, Jasper A. Vrugt. 2026-10-16. Graph-Learned Spring SST Evolution and Land-Ocean Synergy Improve Yangtze Summer Rainfall Prediction. https://arxiv.org/abs/2610.23094

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Atmospheric Chemistry-Climate Modeling with an AI Foundation Model

Atmospheric composition and chemistry-climate interactions are central to predicting future climate change. Yet simulating chemical and aerosol processes is typically the most expensive component of an atmospheric model and is often highly simplified or prescribed entirely. AI weather forecasting foundation models have shown skill rivalling process-based models and are increasingly fine-tuned for new atmospheric tasks. Here, we present the first foundation model to capture interactions between emissions, atmospheric composition, and radiation. We fine-tune the Aurora Earth system foundation model on simulations from the chemistry-climate model UKCA. Our model reproduces daily meteorology, composition, aerosol, and radiation fields of the coupled system with 23 of 27 predicted fields achieving $R^2$ above 0.85 in both the troposphere and stratosphere. It reproduces ozone, CO, and temperature with annual mean biases below 4% and the downwelling shortwave and longwave fluxes to within 6%. Our model remains stable in autoregressive simulation over 50 years while running ~$100\times$ faster than UKCA. By training on paired emissions scenarios, our model learns responses to perturbations by capturing tropospheric ozone's response to changes in NO and CO emissions and surface radiation's response to aerosol. We further show that incorporating MLS satellite ozone retrievals into training corrects biases in the underlying UKCA fields, reducing zonal mean ozone error against collocated observations by 23% between 250 and 20 hPa. Our approach does not suggest process-based modeling can be replaced but demonstrates that a foundation model can be adapted to respond to forcings and learn from observations at a relatively small computational cost.

physics.ao-ph↗
arXiv2026

High-level, efficient ocean modeling at all scales with Oceananigans

We describe the user interface, governing equations, and numerical methods underpinning the ocean modeling software called "Oceananigans". Oceananigans development has been led by the Climate Modeling Alliance as part of a larger effort to build a trainable climate model with quantifiable uncertainty. Notably, Oceananigans is written in the Julia programming language, is optimized for GPUs, and can efficiently simulate all scales of ocean motion ranging from millimeter-scale turbulence in a small box to planetary-scale ocean circulation. We argue that this combination of capabilities is enabled by a design that combines a basic, easy-to-optimize, GPU-friendly structured finite volume algorithm exposed behind a high-level, programmable user interface. This design negotiates a dual mandate for highest-possible performance (to support state-of-the-art applications) and enhanced accessibility (to facilitate adoption and development). The dual mandate aims ultimately to accelerate the progress of Earth system science.

physics.ao-ph↗
arXiv2026

Rescene: stable century-long climate emulation from a frozen weather model

Machine-learning weather models now match or exceed operational numerical prediction in the medium range, but when they are run freely for years they drift, blow up or lose their variability, and dedicated climate emulators are expensive to train. Here we present Rescene, a 0.4 M-parameter wrapper that turns a frozen 1.5$^\circ$, 6-hourly, 69-variable weather model into a stochastic climate emulator. A deterministic slow clock anchors the frozen model to a calendar climatology, and a band-limited AR(1) perturbation, whose spectral envelope is fixed to the observed anomaly spectrum, restores the weather that anchoring removes. A 100-year, 8-member Rescene run shows no detectable drift ($-0.004 \pm 0.014$ K per 100 years), and on the 2022 test year the 16-member ensemble is calibrated (spread/skill 0.89, 0.97 and 0.93 for Z500, T2m and MSLP) with an extratropical synoptic amplitude of 97% of ERA5 on the tuning year. The free run reproduces the geography of day-to-day variability (pattern correlation 0.83 for Z500) and 71% of the observed blocking frequency, where the deterministic wrapper alone has none. Interventions on the frozen operator show a scale-local, approximately linear response in which Z500 perturbations grow upscale 6--14 times more than downscale per 6 h. All constants are fitted on a training year against a rule declared beforehand, and we show that common evaluation shortcuts would have overstated variability and skill. Rescene makes an existing weather model usable for century-long ensembles at negligible training cost.

physics.ao-ph↗