This paper proposes a scalable and efficient architecture for storing, managing, and processing high-density LiDAR point cloud data using spatial database systems integrated with Jupyter-based analytical workflows. LiDAR datasets are inherently large, complex, and multidimensional, often containing billions of points with associated attributes such as intensity, return number, and classification labels. These characteristics impose significant challenges in terms of data storage, indexing, and real-time querying. The proposed approach leverages the capabilities of PostgreSQL / PostGIS as a core relational spatial database environment to enable structured storage and advanced spatial querying of LiDAR point clouds. A data model based on tiled storage and spatial indexing (e.g., GiST and BRIN indexes) is implemented to optimize query performance and support efficient data retrieval at multiple spatial resolutions. In parallel, the architecture supports integration with NoSQL paradigms for distributed storage scenarios, addressing scalability requirements for very large datasets. Processing workflows are executed within interactive environments such as Jupyter Notebook, using Python-based libraries for point cloud manipulation, filtering, and feature extraction. This combination allows seamless interaction between database-level operations (e.g., spatial filtering, aggregation) and in-memory analytical processing, significantly reducing data transfer overhead and improving computational efficiency. The proposed architecture demonstrates that coupling spatial database processing with reproducible notebook-based analytics provides a robust framework for large-scale LiDAR data management, enabling efficient querying, scalable processing, and integration with advanced geospatial and machine learning workflows.

Lidar, NoSQL, Spatial databases, spatial processing, PostGIS

environmental engineering

Presentation: oral presentation