Basing LiDAR surveys on (PostgreSQL) geospatial databases: is it possible?
LiDAR surveys represent one of the largest datasets involved in GIS, corresponding to "point clouds" containing up to several billions of points including more information than simple geospatial coordinates (flight direction, number of returns, color, etc.). The choice of the technology for the management of these data is therefore crucial.
Many tools have been successfully developed to work with LiDAR surveys, based directly on the "LAS files" that collect the point cloud ordered as it is detected: this represent the standard working procedure, even if file readings represent a bottleneck.
In this talk I would like to present an alternative strategy of work, basing LiDAR surveys on a geospatial database. One of the major concerns is if a database is able to manage a so large amount of complex data, such as geospatial ones.
Here I will consider the open source PostgreSQL relational DBMS and its geospatial extensions (PostGIS and PointCloud), showing how points indexing can be crucial for this purpose, and how different types can be used depending on the kind of operation that we have to do. Finally, I will present some simple (but also frequently used in GIS) examples based on kNN searches and inclusions into a given bounding box, to understand which are the achievable performances with PostgreSQL and also to highlight its limits.