From Planning Question to Spatial Database: A Geospatial Data Scheme-Aware Importer for Urban Digital Twins

Abstract

Urban Digital Twins (UDTs) promise cross-sector, evidence-based planning support, yet a foundational challenge persists: ingesting heterogeneous, multi-source geospatial data into a unified, queryable spatial database. This paper presents a manual methodology for translating spatial planning questions and an open-source web-based ingestion interface for configuring spatial databases. The approach begins by formalizing the planning question as a Directed Acyclic Graph (DAG), in which each node represents a measurable concept along with its associated data attributes. These nodes are then translated directly into Django ORM (Object-Relational Mapping) models within a PostgreSQL/PostGIS database, creating a schema-aware foundation for data ingestion. Built on this foundation, a four-tier web application comprising a GeoDjango backend, a React frontend, a Django REST API, and a TiTiler raster tile server was designed to enable non-expert users to upload, field-map, and validate vector, raster, and tabular geospatial data through a single interface. Coordinate reference system reprojection, schema verification, and validation are handled automatically, facilitating the configuration of spatial databases. This model-driven, ORM-based ingestion workflow aims to reduce schema management complexity, compared to manually specified SQL-based approaches, while keeping the application and database layers in sync. Future work will need to evaluate the usability and time reduction of the current tool, and extend the platform to support scenario simulation, LLM-assisted data retrieval from external sources, and real-time connectivity with Smart City sensors via a unified database.