{
  "abstract": "Uncertainty in the terrestrial carbon cycle remains a major constraint in climate projections, partly due to imperfect land surface representations in Earth system models. We propose a data-driven framework for high-resolution land-use reconstruction, treated as a semantic-segmentation task, that combines coarse-resolution scenario data and climate classification with static geophysical features to produce annual land-use maps at 1 km resolution. We introduce TerraNostra-Bench, a change-aware evaluation protocol that separates spatial downscaling from autoregressive temporal rollout and scores models on the rare land-use-change pixels rather than the dominant static background. Using it, we compare convolutional (U-Net) and transformer (Swin) backbones in autoregressive and non-autoregressive schemes against persistence, linear, and shallow convolutional baselines. The autoregressive prior is the most informative input yet accumulates error over long rollouts: autoregressive models lose 5.7-7.2 accuracy points over 13 years, against 2.2-2.8 for non-autoregressive models. This trade-off is governed by the autoregressive choice rather than the backbone. It can be tuned by an inference-time prior-trust threshold that requires no retraining and by the training-time weighting of reconstruction and change objectives; the two levers trace complementary frontiers, with calibration recovering up to 10 accuracy points while keeping change sensitivity above the persistence floor. Extrapolating beyond the training window, backward to 1955-1959 and forward to 2015-2019 under scenario forcing, the models retain most of their in-window skill and show the same trade-off, indicating it is intrinsic to the design. This work provides the methodological foundation and evaluation surface for a forthcoming temporally continuous land-use product.",
  "abstract_shown": true,
  "affiliations": [
    "Barcelona Supercomputing Center",
    "Universitat de València"
  ],
  "article_type": null,
  "arxiv": "2610.27793",
  "author_total": 9,
  "authors": [
    {
      "affs": [
        0
      ],
      "name": "Amirpasha Mozaffari",
      "orcid": "0000-0001-6719-0425"
    },
    {
      "name": "Marina Castaño"
    },
    {
      "affs": [
        0
      ],
      "name": "Stefano MATERIA",
      "orcid": "0000-0001-5635-2847"
    },
    {
      "affs": [
        1
      ],
      "name": "Joan Vedrí",
      "orcid": "0009-0008-5890-1051"
    },
    {
      "affs": [
        0
      ],
      "name": "Dario Garcia-Gasulla",
      "orcid": "0000-0001-6732-5641"
    },
    {
      "affs": [
        0
      ],
      "name": "Etienne Tourigny",
      "orcid": "0000-0003-4628-1461"
    },
    {
      "affs": [
        0
      ],
      "name": "Jordi Varela-Agrelo",
      "orcid": "0009-0009-8707-155X"
    },
    {
      "affs": [
        0
      ],
      "name": "Mario C. Acosta",
      "orcid": "0000-0001-7054-8168"
    },
    {
      "affs": [
        0
      ],
      "name": "Amanda Cardoso Duarte",
      "orcid": "0000-0002-9340-958X"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/4539xjysnqqn2.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.27793",
  "fulltext_url": null,
  "id": "4539xjysnqqn2",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.27793",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "physics.ao-ph"
    ],
    [
      "openalex",
      "2306|Land Use and Ecosystem Services"
    ]
  ],
  "openalex": "W7223133573",
  "pdf_url": "https://arxiv.org/pdf/2610.27793",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-20",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/4539xjysnqqn2.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:34:04Z",
      "native_id": "2610.27793",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.27793"
    },
    {
      "first_seen": "2026-11-03T12:20:16Z",
      "native_id": "23768256",
      "source": "zenodo",
      "url": "https://zenodo.org/records/23768256"
    }
  ],
  "subjects": [
    {
      "field": "earth",
      "origin": "arxiv:physics.ao-ph",
      "subfield": "atmospheric"
    }
  ],
  "title": "TerraNostra-Bench: Benchmarking Spatial Downscaling and Temporal Rollout for Global 1-km Land-Use Reconstruction",
  "title_text": "TerraNostra-Bench: Benchmarking Spatial Downscaling and Temporal Rollout for Global 1-km Land-Use Reconstruction",
  "type": "preprint",
  "updated_at": "2026-11-03T12:20:16Z",
  "url": "https://research.draconitic.com/paper/4539xjysnqqn2",
  "venue": "arXiv",
  "versions": []
}