Alex Burnham - Keynote: Lowering the barrier of entry to spatiotemporal data analysis

Learn about SpaceTime, a Python/R toolkit that simplifies spatiotemporal data analysis for non-GIS experts, featuring efficient data alignment and cube operations.

Key takeaways
  • SpaceTime is a Python/R toolkit aimed at lowering barriers for spatiotemporal data analysis, particularly for researchers in life sciences and social sciences who aren’t GIS experts

  • Key capabilities include:

    • Reading common raster data types (TIFFs, NC4, PNGs, etc.)
    • Aligning datasets with different spatial reference systems and grid sizes
    • Trimming datasets to common bounding boxes
    • Scaling time dimensions (e.g., daily to monthly)
    • Mathematical operations between cubes (“cube smasher”)
    • Converting to/from tabular formats
  • The workflow has two main components:

    • Pre-cube operations: cleaning and aligning raw files
    • Post-cube operations: manipulating clean spatiotemporal cubes
  • Built on Google Earth Engine with focus on user-friendly defaults while maintaining power/flexibility

  • Reduces code complexity compared to alternatives (fewer lines of code needed) while maintaining computational efficiency

  • Current limitations/future work:

    • Working on distributed computing capabilities
    • Building open source community
    • Improving documentation
    • Adding network/connectivity analysis features
    • Expanding beyond raster data
  • Ideal for ecological/environmental applications like:

    • Climate change analysis
    • Species distribution modeling
    • Agricultural modeling
    • Weather data analysis
  • Available on PyPI (Python) and GitHub (R version), with ongoing development prioritizing the Python version

  • Focus on academic/research use cases but potential for commercial applications

  • Emphasizes compatibility with other tools/formats while simplifying initial data preparation steps

  • Development supported by NSF EPSCoR grant through Barracuda project collaboration between UVM and UMaine