km Scales Hacked at Global Scale

km-scale climate models promise unprecedented insights into fine-scale processes, but their massive data volumes and heterogeneous formats pose critical challenges for analysis or even multi-model intercomparison. We addressed these barriers through a global hackathon involving 600+ participants across 10 nodes who collaboratively analyzed outputs from diverse km-scale regional and global climate models, largely from the DYAMOND 3 intercomparison.

We enabled the intercomparison by standardizing all datasets to a common HEALPix grid, providing them as cloud-accessible Zarr stores indexed with Intake and deploying a unified Python environment via JupyterHub at the hackathon nodes. This infrastructure avoided the download-and-scan pattern common with large NetCDF collections, enabling faster interactive workflows.

Concise tutorials and this infrastructure enabled all participating teams—regardless of background or resources—to interactively explore km‑scale features such as extreme precipitation, mesoscale organization, and fine‑scale ocean–atmosphere coupling across models.

We present the technical workflow and lessons learned from rapidly deploying this infrastructure across distributed nodes and invite the community to explore these openly accessible datasets at https://digital-earths-global-hackathon.github.io/catalog .