2

I'm debating installing the PostGIS extension for Postgres. However, upon looking at the types in Postgres itself, it seems to support the basic Geometric types, including:

8.8.1. Points
8.8.2. Lines
8.8.3. Line Segments
8.8.4. Boxes
8.8.5. Paths
8.8.6. Polygons
8.8.7. Circles

So then, what is the use case for using PostGIS and what would that give me that Postgres itself wouldn't give me when working with spatial data? (Ignoring UI considerations).

David542
  • 104,438
  • 178
  • 489
  • 842
  • 3
    Well, in a nutshell, there's a difference between 2D geometry happening in a plane and geospatial geometry happening on an ellipsoid (that might be neglectable in some cases in others not so much...). The "g" in "GIS" stands for "geo(graph(y|ic)?)", so the latter. – sticky bit Dec 07 '21 at 01:47
  • Geospatial calculations get complicated very fast. (Search for 'haversine formula'). And that is for the simple operation of point-to-point distance. Postgis supplies a complete set of Geospatial functions; Postgres itself does not contain those functions. – Belayer Dec 07 '21 at 04:42
  • @Belayer I see. Why doesn't postgres have any of the `ST_` functions? It seems just to have some common 'shapes' that can be worked with. – David542 Dec 07 '21 at 04:53
  • 3
    If you do need geo-**spatial** calculations then there is no way around PostGIS. If you only need to store (geometric) shapes, then core Postgres will be enough. Operations on geometric shapes is something different than operations on geospatial objects. –  Dec 07 '21 at 06:59

2 Answers2

4

It would all depend on your use case. If you're planing to apply simple spatial operations, such as containment or distance, the native geometric data types would do the trick. But here are a few things that speak for PostGIS anyway:

  • PostGIS is a huge open source project in constant development.
  • it has over 1.3k really kickass functions that abstract most of the cumbersome math involved in spatio-temporal queries. Think only about operations like Spatial Reference System (SRS) transformations, projections or simply handling polyhedral surfaces and you will soon realize the complexity involved with geo spatial data. Believe me, it is not trivial. Even the simple feature of rendering geo spatial data in different formats like WKT, KML or GeoJSON can save you a lot of time!
  • it enables you to easily access the data from most of modern GI Software. Feel like plotting your data into a map? ;)
  • it can serve as data source for OGC Standard web services, such as Web Feature Service (WFS) or Web Map Service (WMS). Take a look at GeoServer.
  • it is pretty well adopted in the GIS community, which means you will find a vast amount of content and support online.
  • it supports both vector and raster data types.
  • it offers (multidimensional) spatial indexing. Geospatial data sets are known to be huge, so a proper indexing mechanism is imperative to perform spatial operations in a decent query time.
  • it offers network topology support.
  • it is the building block of other pretty handy extensions, e.g. pgRouting and ogrfdw.

The list of advantages goes on and on.

Here a few real life use cases that can be easily addressed with PostGIS:

Jim Jones
  • 18,404
  • 3
  • 35
  • 44
1

The Earth's surface is not an infinite 2D plane. How reasonable it is to approximate it as one depends on what you are doing.

jjanes
  • 37,812
  • 5
  • 27
  • 34