For over 40 years satellites have been collecting (and weather and climate models have been generating) petabytes of observations and estimates of properties of the Earth. Despite being public data, these data are relatively underutilized due to limitations in the standard approach of downloading raw data locally so that it can be processed using specialized geospatial software.

While the Jupyter architecture has proven to be valuable for documenting complex analyses in a reproducible manner, its use for non-geospatial data analysis has been relatively limited. This session will show how several key technologies can be combined to facilitate large-scale geospatial data analysis:

• Google Earth Engine - A cloud-based geospatial analysis platform that provides access to petabytes of Earth science data via a Python API.
• IPyLeaflet - a Jupyter widget that can add geospatial data to an interactive map.
• Jupyter Notebooks - An open-source, web-based environment that supports documents that combine code and computational results with text narrative, mathematics, images, and other media. These notebooks provide an environment for interactive exploration of data and development of well documented algorithms.

The advantages of this technology pairing will be demonstrated by showing how landcover change can be estimated from satellite imagery over entire countries.