{"paper":{"title":"Learning to Advect: A Neural Semi-Lagrangian Architecture for Weather Forecasting","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A neural semi-Lagrangian architecture decomposes weather forecasting into advection, diffusion, and reaction blocks on latent variables.","cross_cats":["physics.ao-ph"],"primary_cat":"cs.LG","authors_text":"Carlos A. Pereira, Christopher Subich, David Millard, Eldad Haber, Emilia Diaconescu, Raymond J. Spiteri, Sasa Zhang, Shoyon Panday, Siddharth Rout, Siqi Wei, St\\'ephane Gaudreault, Valentin Dallerit","submitted_at":"2026-01-29T01:20:21Z","abstract_excerpt":"Recent machine-learning approaches to weather forecasting often employ a monolithic architecture in which distinct physical mechanisms-advection (long-range transport), diffusion-like mixing, thermodynamic processes, and forcing-are represented implicitly within a single large network. This is particularly problematic for advection, where long-range transport typically requires expensive global interaction mechanisms or deep stacks of local convolutional layers. To mitigate this, we present PARADIS, a physics-inspired global weather prediction model that enforces inductive biases on network be"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Evaluated on ERA5 benchmarks, PARADIS achieves competitive deterministic forecast skill, with particularly strong short-lead performance, while preserving substantially better spectral fidelity and forecast activity during medium-range rollouts.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That imposing a functional decomposition into advection, diffusion-like mixing, and reaction blocks on latent variables will produce physically meaningful trajectories and superior spectral properties without introducing artifacts from the differentiable interpolation or learned latent modes.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"PARADIS decomposes weather dynamics into advection via differentiable Neural Semi-Lagrangian transport, depthwise-separable diffusion, and pointwise reaction terms, yielding competitive ERA5 forecasts with improved spectral fidelity in medium-range rollouts.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A neural semi-Lagrangian architecture decomposes weather forecasting into advection, diffusion, and reaction blocks on latent variables.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"74a2f766e5f6000330b48131db20f6ea258a86f2ac95c1b46847ce782e4d89b3"},"source":{"id":"2601.21151","kind":"arxiv","version":2},"verdict":{"id":"4c7d2413-f17c-41a3-a0fd-6a188bbcb19e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T10:12:27.798996Z","strongest_claim":"Evaluated on ERA5 benchmarks, PARADIS achieves competitive deterministic forecast skill, with particularly strong short-lead performance, while preserving substantially better spectral fidelity and forecast activity during medium-range rollouts.","one_line_summary":"PARADIS decomposes weather dynamics into advection via differentiable Neural Semi-Lagrangian transport, depthwise-separable diffusion, and pointwise reaction terms, yielding competitive ERA5 forecasts with improved spectral fidelity in medium-range rollouts.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That imposing a functional decomposition into advection, diffusion-like mixing, and reaction blocks on latent variables will produce physically meaningful trajectories and superior spectral properties without introducing artifacts from the differentiable interpolation or learned latent modes.","pith_extraction_headline":"A neural semi-Lagrangian architecture decomposes weather forecasting into advection, diffusion, and reaction blocks on latent variables."},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"bf44987bb848d6b73e191961527e71536ac80f4e617ee526fdb57dcf1b8cd3a3"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}