ODINN.jl - Multi-language Geoscientific Machine Learning for Glacier Evolution Modeling

Explore a 10-minute conference talk from JuliaCon 2024 introducing ODINN.jl, a global glacier evolution model that combines scientific machine learning and multi-language capabilities in Julia. Discover how this innovative framework handles Universal Differential Equations for glacier ice flow, solving hundreds of thousands of ODEs in parallel. Learn about the model's ability to perform functional inversions to uncover physical laws of geophysical processes while assimilating heterogeneous, noisy, and sparse observations. Understand how ODINN.jl leverages Julia's SciML differentiable programming capabilities and integrates with Python tools like PyCall.jl, the Open Global Glacier Model, and Xarray for data processing and management. Gain insights into the future of geoscientific modeling, where flexible frameworks combine mechanistic models based on domain knowledge with machine learning models that learn from massive Earth observations. See how Julia's multi-language capabilities enable smooth interoperability between new tools and established frameworks, accelerating the adoption of new methods in existing scientific communities.