Workflow automation of ICON model runs using auto-icon

Abstract

Weather and climate modeling workflows comprise not only the numerical simulation itself, but also extensive pre-processing, data management, and post-processing tasks. Many of these tasks are repetitive across experiments and are therefore well suited for automation. Despite this potential, no widely adopted workflow management solution currently exists within the community of the ICOsahedral Nonhydrostatic (ICON) weather and climate model. To address this gap, auto-icon was developed as an extension to the workflow manager Autosubmit, enabling the orchestration and execution of ICON-based modeling workflows. In this work, we show several representative use cases that demonstrate the applicability of auto-icon to a range of modeling scenarios. The presented applications highlight improvements in workflow throughput, usability, reproducibility, and sustainability achieved through automated workflow management. More generally, this work illustrates the broader benefits that workflow management systems can provide for scientific modeling and computational research.