HPC-oriented Canonical Workflows for Machine Learning Applications in Climate and Weather Prediction

Abstract Machine learning (ML) applications in weather and climate are gaining momentum as big data the immense increase High-performance computing (HPC) power paving way. Ensuring FAIR reproducible ML practices significant challenges for Earth system researchers. Even though principle is well known to many scientists, research communities slow adopt them. Canonical Workflow Framework Research (CWFR) provides a platform ensure FAIRness reproducibility of these without overwhelming This conceptual paper envisions holistic CWFR approach towards climate, focusing on HPC data. Specifically, we discuss Fair Digital Object (FDO) (RO) DeepRain project achieve granular reproducibility. that aims improve precipitation forecast Germany by using ML. Our concept envisages raster datacube provide harmonization fast scalable access. We suggest Juypter notebook single experiment. In addition, envision JuypterHub distributed central connects all elements resources researchers via an easy-to-use graphical interface.