GNN-VIEW: Visual Inspection and Explanation of Graph-based Earth-System Models
Abstract
Machine learning Earth-system models are becoming increasingly important, and so is their development. Graph-based models, one of the most prominent approaches, allow developers to constrain the learning process before training via graph connectivity. Yet, developing these models remains challenging: the graph is built algorithmically with little intuition about the final result beyond static plotting, which requires extra work and is often not detailed enough. Even after successful training, how the model arrives at forecasts remains obscure. We present GNN-VIEW, an interactive browser-based tool that supports the development of such systems. GNN-VIEW exposes the graph's information pathways, allowing model developers to debug the graph structure before training. After training, GNN-VIEW uses attribution-based explainability to reveal which input features are used by the trained model for a given forecast. This helps developers understand why a model fails. We showcase the usefulness of GNN-VIEW in the construction stage of a graph for an ocean forecasting model, and in explaining poorly predicted precipitation of a trained weather forecasting model during Hurricane Laura.
Related
- YAC v3.19.0: scalable online interpolation and flexible coupling for Earth system models
- The calibration deficit: a challenge for Earth system simulation and emulation
- How to Use Score-Based Diffusion in Earth System Science: A Satellite Nowcasting Example
- The ocean: a key Earth system regulator with boundaries.
- Generalizable Neural Downscaling of Earth System Model Wind Fields via Continuous Dynamics Modeling
BibTeX
@misc{lucatelli2026gnnview,
title = {{GNN-VIEW: Visual Inspection and Explanation of Graph-based Earth-System Models}},
author = {Alessandro Lucatelli and Ferdinand Unterhuber and Katharina Krenzer and Mennatallah El-Assady and Benedikt Soja and Christina Humer and Joel Oskarsson},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.35693},
eprint = {2610.35693},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.35693},
}
RIS
TY - UNPB
TI - GNN-VIEW: Visual Inspection and Explanation of Graph-based Earth-System Models
AU - Alessandro Lucatelli
AU - Ferdinand Unterhuber
AU - Katharina Krenzer
AU - Mennatallah El-Assady
AU - Benedikt Soja
AU - Christina Humer
AU - Joel Oskarsson
PY - 2026
DA - 2026/10/26
PB - arXiv
DO - 10.48550/arxiv.2610.35693
UR - https://arxiv.org/abs/2610.35693
AB - Machine learning Earth-system models are becoming increasingly important, and so is their development. Graph-based models, one of the most prominent approaches, allow developers to constrain the learning process before training via graph connectivity. Yet, developing these models remains challenging: the graph is built algorithmically with little intuition about the final result beyond static plotting, which requires extra work and is often not detailed enough. Even after successful training, how the model arrives at forecasts remains obscure. We present GNN-VIEW, an interactive browser-based tool that supports the development of such systems. GNN-VIEW exposes the graph's information pathways, allowing model developers to debug the graph structure before training. After training, GNN-VIEW uses attribution-based explainability to reveal which input features are used by the trained model for a given forecast. This helps developers understand why a model fails. We showcase the usefulness of GNN-VIEW in the construction stage of a graph for an ocean forecasting model, and in explaining poorly predicted precipitation of a trained weather forecasting model during Hurricane Laura.
ER -