Tom Vo - xCDAT (Xarray Climate Data Analysis Tools) | SciPy 2024

XCDAT is a Python package for climate data analysis that combines X-Ray capabilities with features from CDAT (Climate Data Analysis Tools), focusing on routine climate analysis operations

Core capabilities include:

• File I/O for NetCDF, Iris, OpenNAP formats
• Spatial and temporal averaging
• Regridding (horizontal and vertical)
• Climatology calculations
• Anomaly analysis
• Bounds handling for robust calculations
• Parallel processing with Dask integration

The package emphasizes:

• Clean, intuitive APIs
• Software sustainability
• Reproducible science
• CF-compliant metadata handling
• Interoperability with other scientific Python tools

Key design advantages:

• Single-line operations vs multi-line X-Ray equivalents
• Automatic metadata interpretation
• Built-in parallelization
• Proper chunking for large datasets
• Weight-based calculations for accurate regional averaging

Development approach:

• Feedback-driven development with climate scientists
• Open source community focus
• Funded through Numfocus for long-term stability
• Part of Pangeo ecosystem
• Built on modern technology stack

Designed for growing challenges:

• Increasing model resolution
• Larger dataset sizes
• Need for efficient parallel processing
• Cross-dataset comparisons
• Ensemble analysis