NetCDFaster: Optimizing NetCDF data querying and geo-visualization using high-performance machine learning

Extracting and visualizing multidimensional subsets from large NetCDF datasets is often repetitive, time-consuming, and technically demanding. Existing tools typically trade off between performance and usability, either optimizing data retrieval at the cost of user experience or providing user-friendly interfaces that struggle with performance issues. NetCDFaster bridges this gap by integrating a lightweight CatBoost classifier model into a web-based platform that automatically selects optimal data retrieval strategies. The benchmark results demonstrate that NetCDFaster reduces end-to-end retrieval times by more than 80%, allowing researchers to efficiently explore data subsets and visualize results. This significantly enhances the speed and effectiveness of scientific workflows involving NetCDF data.