Deck et al. (2026) ClimaLand: A Land Surface Model Designed to Enable Data‐Driven Parameterizations

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Journal of Advances in Modeling Earth Systems
• Year: 2026
• Date: 2026-01-01
• Authors: Katherine M. Deck, Renato K. Braghiere, Alexandre A. Renchon, Julia Sloan, Gabriele Bozzola, Edward Speer, J. Ben Mackay, Teja Reddy, Kevin Phan, Anna L. Gagné‐Landmann, Yuchen Li, Dennis Yatunin, Andrew Charbonneau, Nat Efrat‐Henrici, Eviatar Bach, Shuang Ma, Pierre Gentine, Christian Frankenberg, A. Anthony Bloom, Yujie Wang, Marcos Longo, Tapio Schneider
• DOI: 10.1029/2025ms005118

Research Groups

Not specified in abstract.

Short Summary

This paper introduces ClimaLand, a new land surface model (LSM) designed to overcome limitations in sub-grid parameterization, calibration, and uncertainty quantification in existing LSMs. It demonstrates ClimaLand's computational efficiency via GPU leverage and its modular architecture, which facilitates integration with machine learning libraries.

Objective

• To present the first version of ClimaLand, a new land surface model, designed to address challenges in parameterizing sub-grid processes, calibration, uncertainty quantification, and adaptability to machine learning parameterizations within climate models.

Study Configuration

• Spatial Scale: 10–100 km (scales relevant to climate models)
• Temporal Scale: Not specified in abstract.

Methodology and Data

• Models used: ClimaLand (new Land Surface Model developed in Julia).
• Data sources: "in situ and satellite data" (mentioned as data for training future ML parameterizations, not explicitly for the validation exercises described in this paper).

Main Results

• ClimaLand effectively leverages graphics processing units (GPUs) to achieve computational efficiency.
• Its modular architecture and implementation in the high-level programming language Julia enable seamless integration with machine learning libraries.
• The model's design facilitates efficient simulation, calibration, and uncertainty quantification.

Contributions

• Presents a novel land surface model, ClimaLand, specifically engineered with a modular architecture and high-level programming language (Julia) to overcome existing LSM limitations regarding sub-grid parameterization, calibration, and uncertainty quantification.
• Introduces an LSM designed for direct integration with modern machine learning parameterizations, enhancing adaptability and future research capabilities.
• Demonstrates the computational efficiency of the new model through GPU utilization.

Funding

Not specified in abstract.

Citation

@article{Deck2026ClimaLand,
  author = {Deck, Katherine M. and Braghiere, Renato K. and Renchon, Alexandre A. and Sloan, Julia and Bozzola, Gabriele and Speer, Edward and Mackay, J. Ben and Reddy, Teja and Phan, Kevin and Gagné‐Landmann, Anna L. and Li, Yuchen and Yatunin, Dennis and Charbonneau, Andrew and Efrat‐Henrici, Nat and Bach, Eviatar and Ma, Shuang and Gentine, Pierre and Frankenberg, Christian and Bloom, A. Anthony and Wang, Yujie and Longo, Marcos and Schneider, Tapio},
  title = {ClimaLand: A Land Surface Model Designed to Enable Data‐Driven Parameterizations},
  journal = {Journal of Advances in Modeling Earth Systems},
  year = {2026},
  doi = {10.1029/2025ms005118},
  url = {https://doi.org/10.1029/2025ms005118}
}