AI/ML and the Future of Earth System Science Technical Workforce: How NSF Unidata is Preparing the next generation of ML Professionals

NSF Unidata recognizes the growing importance of machine learning (ML) skills in the Earth Systems sciences. Despite the transformative potential of ML techniques, these skills are not routinely taught to undergraduate students. To bridge this gap and equip the next generation of scientists, NSF Unidata has launched a comprehensive outreach initiative focused on fostering ML literacy for Earth Systems Science. This initiative includes an NSF-supported CyberTraining program in partnership with Metropolitan State University of Denver, a blog series, Jupyter notebooks, instructor-led training, and one-on-one mentorship.

The NSF-supported CyberTraining program is designed to equip Earth Systems scientists, students, and educators with essential machine learning skills and knowledge to leverage cyberinfrastructure tools in their research and professional careers. The program's primary focus is on delivering scalable, relevant machine learning experiences and foundational content for upper-level undergraduate and graduate curricula in Earth Systems Science. It bridges the gap between theoretical understanding of ML algorithms and their practical, "low-code" implementations in Earth Systems research. Participants learn to assess the suitability of various machine learning techniques for Earth Systems problems, understand the intricacies of data preparation, model selection, and evaluation, and gain hands-on experience in applying ML to real-world Earth Systems scenarios. The program emphasizes the long-term impact of these machine learning skills on career advancement in Earth Systems Science, ensuring participants are well-prepared for the increasing integration of ML in this domain.

NSF Unidata's machine learning blog series highlights applications of machine learning in atmospheric, and geological research. These blogs showcase real-world examples of how ML is utilized in weather prediction, climate modeling, and data driven analysis. These blogs aim to raise the curiosity of undergraduate students and motivate them to explore ML further for their own applications and research questions.

NSF Unidata has also curated a collection of open-source Jupyter notebooks specifically designed for Earth Systems science students. These interactive notebooks serve as self-paced tutorials, guiding students through machine learning case studies using Python. Some of these notebooks are further refined into Project Pythia Cookbooks that provide more of an in-depth tutorial into specific datasets and workflows. From data preprocessing and exploratory analysis to building and evaluating ML models, these notebooks cover essential concepts and techniques relevant to atmospheric research.

NSF Unidata has also established a one-on-one virtual office hours program to provide personalized support and guidance. This program connects undergraduate and graduate students, as well as faculty, with AI/ML expertise at NSF Unidata.

NSF Unidata's multi-faceted outreach approach aims to democratize machine learning education in the Earth Systems sciences. By equipping undergraduate and graduate Earth Systems Science students with essential ML skills and inspiring them to embrace data-driven approaches, NSF Unidata is actively shaping the future of Earth Systems Science research and industrial workforce.