Question

I am interested in best practices for designing relational DB tables for holding spatio-temporal data. Specifically data that will be kept in such tables are custom geometries that have certain validity period, geometry definition as well as hierarchical aspect(certain geometries will be children of other geometries).

I curious if someone can point me to a good material on this subject or could suggest specific implementation.

Thank you.

Answer 1

I’d use PostGIS (http://postgis.refractions.net/) for geometry type and make a table like this:

CREATE TABLE data (
    geometry geometry,
  valid_from timestamp,
  valid_till timestamp,
  check(valid_till >= valid_from)
);

PostGIS can make spatial queries, so you can query database for all geometries in a specific geometry (e.g. query for all geometries in a geometry representing a State, or a County).

To get the validity period, you should add to this query additional condition for getting only the rows where (valid_from >= now() and valid_till <= now()).

You will also need indexes on all three columns of course. On the geometry column there should be a spatial index.

All the information about the spatial queries and geometry type and geometry indexes you can find on the PostGIS site.