{"paper":{"title":"Reduced-Order Surrogates for Forced Flexible Mesh Coastal-Ocean Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Koopman autoencoders with forcings deliver accurate year-long reduced-order surrogates for coastal-ocean models.","cross_cats":["cs.AI","cs.LG","physics.ao-ph","physics.flu-dyn"],"primary_cat":"cs.CE","authors_text":"Allan P. Engsig-Karup, Freja H{\\o}gholm Petersen, Jesper Sandvig Mariegaard, Rocco Palmitessa","submitted_at":"2026-02-05T07:59:58Z","abstract_excerpt":"While proper orthogonal decomposition (POD)-based surrogates are widely explored for hydrodynamic applications, the use of Koopman autoencoders for real-world coastal-ocean modelling remains relatively limited. This paper introduces a flexible Koopman autoencoder formulation that incorporates meteorological forcings and boundary conditions, and systematically compares its performance against POD-based surrogates. The Koopman autoencoder employs a learned linear temporal operator in latent space, enabling eigenvalue regularization to promote temporal stability. This strategy is evaluated alongs"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Across all cases, the reduced order surrogates with temporal unrolling achieve high accuracy with relative root-mean-squared-errors of 0.0068-0.14 and R²-values of 0.61-0.995... In two of the three cases, the Koopman Autoencoder have higher accuracy than the POD-based surrogates. Comparing to in-situ observations, the surrogate yields -0.64% to 12% increase in water surface elevation prediction error when compared to prediction errors of the physics-based model.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the learned linear temporal operator in latent space, regularized for eigenvalue stability, continues to produce bounded long-term trajectories when driven by real meteorological forcings outside the three tested regimes.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Koopman autoencoders with forcings and temporal unrolling deliver accurate year-long predictions for coastal-ocean models at 300-1400x speedup, outperforming POD in two of three cases.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Koopman autoencoders with forcings deliver accurate year-long reduced-order surrogates for coastal-ocean models.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"26688f69c324a3d064032241221bd00d249dafca8bde18d2c3037f2c76046497"},"source":{"id":"2602.05416","kind":"arxiv","version":2},"verdict":{"id":"e8f937db-1547-430b-baf0-d4b4ac087ad3","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T07:28:27.104955Z","strongest_claim":"Across all cases, the reduced order surrogates with temporal unrolling achieve high accuracy with relative root-mean-squared-errors of 0.0068-0.14 and R²-values of 0.61-0.995... In two of the three cases, the Koopman Autoencoder have higher accuracy than the POD-based surrogates. Comparing to in-situ observations, the surrogate yields -0.64% to 12% increase in water surface elevation prediction error when compared to prediction errors of the physics-based model.","one_line_summary":"Koopman autoencoders with forcings and temporal unrolling deliver accurate year-long predictions for coastal-ocean models at 300-1400x speedup, outperforming POD in two of three cases.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the learned linear temporal operator in latent space, regularized for eigenvalue stability, continues to produce bounded long-term trajectories when driven by real meteorological forcings outside the three tested regimes.","pith_extraction_headline":"Koopman autoencoders with forcings deliver accurate year-long reduced-order surrogates for coastal-ocean models."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2602.05416/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":38,"sample":[{"doi":"10.48550/arxiv.2003.02236","year":2003,"title":"doi:10.48550/arXiv.2003.02236","work_id":"9e4dfbc9-d2c3-4763-9a9d-b5155ccbdd64","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1029/2024gl112835","year":null,"title":"Brunton, S.L., Brunton, B.W., Proctor, J.L., Kutz, J.N","work_id":"5b82a8c1-42e8-4054-b98a-fcb41281ec77","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1371/journal.pone.0150171","year":null,"title":"and Brunton, Bingni W","work_id":"035644e0-a7b4-4aac-98f0-13e8ea87ddb2","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"On the Properties of Neural Machine Translation: Encoder-Decoder Approaches","work_id":"a6dc7801-0da7-423e-ba1d-00d246875f48","ref_index":4,"cited_arxiv_id":"1409.1259","is_internal_anchor":true},{"doi":"","year":2025,"title":"URL:https://comune","work_id":"231753d6-445b-446a-8f2e-4b36638189c1","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":38,"snapshot_sha256":"e2db85972efae3c0a7d54749cd122081b6f7430c7ff88cccfc552fa17615ff94","internal_anchors":6},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}