Mentor(s): Robbyn Abbitt

Spatial data pipelines receive, store, transform, and analyze geographic information. Despite decades of building and researching such structures, few models are ‘general’ and which reduce the necessary functions to their minimal extent. Most implementations are, on the contrary, highly specialized, proprietary, and cloud-dependent, designed for specific enterprise use.

This paper presents a minimal implementation of a spatial data pipeline which can ingest heterogeneous spatial data (raster, vector, etc) using only free, open-source components. Its purpose is not to compete with larger commercial products, but to expose the architectural invariants common to all spatial pipelines. The model isolates the minimal set of operations which any spatial data stream must perform, while still successfully ingesting, storing, and transforming spatial data.

Explore the Project