Spatial Data Types and Spatial Reference System

Hui Hu Ph.D.

Department of Epidemiology

College of Public Health and Health Professions & College of Medicine

January 24, 2018

Spatial Data Types and Spatial Reference System

Lab: PostGIS Part 1

Spatial Data Types and Spatial Reference System

Spatial Data Types

  • Geometry:
    - the planar type
    - the very first model and still the most popular type
    - the foundation of the other types
    - uses the Cartesian math
  • Geography:
    - the spheroidal geodetic type
    - lines and ploygons drawn on the earth's curved surface
  • Raster:
    - the multiband cell type
    - space as a grid of rectangular cells, each containing a numeric array of values
  • Topology:
    - the relational model type
    - models the world as a network of connected nodes, edges, and faces
    - network
  • These four types can coexist in the same database:
    - e.g. you can have a geometry that defines the boundaries of a plant, and a raster that defines the concentration of toxic waste along each part of the boundary

Geometry Type

We can represent all geographical entities in 2D using 3 building blocks:




  • Simplified models of reality, and will never perfectly mimic the real thing
  • Geometry type treats the world as a flat Cartesian grid

Geography Type

Similar to geometry, but account for the curvature of the earth

Shape of the earth

  • Surface: The Earth's real surface
  • Ellipsoid: Ideal, smooth surface
  • Geoid: Bumpy surface, where gravity is equal for all locations

Raster Type

Vector Model

- points, lines, polygons
- geometry and geography type

Raster Model
- exhaustive regular or irregular partitioning of space



Topology Type

  • Network of points, lines, and polygons
  • Not concerned with the exact shape and location of geographic features, but with how they're connected to each other
  • Useful in many applications:
    -  parcel (land lot) data, to ensure that the change of one parcel boundary adjusts all other parcels that share that boundary change as well
    -  road management, water boundaries, etc.
    -  architecture

Spatial Reference System

  • SRS is the production of geodetics and cartography
    -  geodetics: the science of measuring and modeling the earth
    -  cartography: the science of representing the earth on flat maps
  • Why do we need SRS?
    -  to bring in data from disparate sources that use different SRSs and be able to overlay one atop another
  • Many standards of SRS:
    -  most common one is the European Petroleum Survey Group (EPSG) numbering system
    -  take any two sources of data with the same EPSG number, and they will overlay perfectly


  • Spatial Reference IDentifier
    -  It defines all the parameters of our data’s geographic coordinate system and projection.
    -  An SRID is convenient because it packs all the information about a map projection (which can be quite complex) into a single number.
  • EPSG is a very recent SRS numbering system
    -  If you are using data from a few decades ago, you won't find EPSG number

  • The constituent pieces that form an SRS:
    -  ellipsoid
    -  datum
    -  projection

Shape of the earth

  • Gauss determined in the early 19th century that the surface of the earth can be defined using gravitational measurements
    -  geoid: where gravity is equal for all locations
  • Geoid is far from spherical
    -  the core of the earth is not homogenous
    -  mass is distributed unevenly

  • Geoid is the foundation of both planar and
    geodetic models


  • Simplifications of the geoid which are generally good enough for most geographic modeling needs
  • An ellipsoid is merely a 3D ellipse
  • Instead of one ellipsoid to rule us all, people on different continents wanted their own ellipsoids to better reflect the regional curvature of the earth
  • Today, the world is settling on the World Geodetic System (WGS 84) and Geodetic Reference System (GRS 80) ellipsoids
    - WGS 84 is the standard of choice, and is what all GPS systems are based on

Common ellipsoids and their ellipsoidal parameters

  • Lon/lat with different ellipsoid are not the same
    -  they use different grounding points
    -  it's important to not just call things lon/lat: you can have NAD27 lon/lat, NAD80 lon/lat, etc. Each will be subtly different


  • Ellipsoid only models the overall shape of the earth
    -  after picking out an ellipsoid, you need to anchor it to use it for real-world navigation
    -  even if two reference systems use the same ellipsoid, they can still have different anchors, or datum, on earth
  • Defines the position of the spheroid relative to the center of the earth.
  • Global datum:
    - uses the earth's center of mass as the origin
  • Local datum:
    - aligns its spheroid to closely fit the earth's surface in a particular area
    - a point on the surface of the spheroid is matched to a particular position on the surface of the earth
    - the coordinate system origin of a local datum is not at the center of the earth

Coordinate Reference System

  • A coordinate reference system is only one necessary ingredient that goes into the making of an SRS and isn't SRS itself
    -  used to identify a point on your reference ellipsoid
  • Most popular coordinate reference system for use is the geographical coordinate system
    -  also known as geodetic coordinate system or simply lon/lat
  • Longitude and latitude
  • Units: Degrees (DMS or DD)


Taking an ellipsoidal earth and squashing it onto a flat surface

  • Projection has distortion built in
    - because geodetic and 3D globes are ellipsoidal, they by definition do not refer to a flat surface
  • Why do we need to have 2D projections?
    -  the mathematical and visual simplicity that comes with planar (Euclidean) geometry


  • How exactly you squash an ellipsoidal earth on a flat surface depends on what you are trying to optimize for
  • In creating a projection, we try to balance four conflicting features:
    -  measurement
    -  shape: how accurately does it represent angles
    -  direction: is north really north
    -  range of area supported
  • E.g. if you want to span a large area, you have to either give up measurement accuracy or deal with the pain of maintaining multiple SRSs and some mechanisms to shift among them

Projection Types

Cylindrical projections 

Conic projections

Azimuthal projections

Orientation of the paper roll around the globe

Main classes of planar coordinate systems

  • Lambert Azimuthal Equal Area (LAEA)
    -  good for measurement and can cover large areas, but not great for shape
    -  US National Atlas (EPSG:2163)
  • Lambert Conformal Conic (LCC)
    -  preserve shape more than area, good for measurement for the regions they serve, and distort poles
    -  best used for middle latitudes with east-west orientation
  • Universal Trans Mercator (UTM)
    -  good for measurement, shape, and direction, but only span six-degree longitudinal strips, cannot be used for the polar regions
  • Mercator
    -  good for preserve shape and direction, and spanning the globe, but not good for measurement
    -  common favorites for web map display since we only need to maintain one SRID
  • National grid systems
    -  variant of UTM or LAEA, but are used to define a restricted region, such as a country
  • State plane
    -  US spatial reference systems, usually designed for a specific state
    -  most are derived from UTM

Universal Transverse Mercator Coordinate System

  • World divided into 60 six-degree-wide zones
  • From 80S to 84N
  • Zones numbered 1-60 (N&S), W to E, starting at 180W


Differences between projections

What spatial reference system is appropriate?

  • Excellent: covers the globe
  • Good: covers a large country like the US; the measurements for the area served are usually within a meter for length, area, and distance calculations
  • Medium: covers several degrees or a large state; measurements are accurate within meters, but can be as much as 10 meters off
  • Bad: measurements don't have useful units

Lab: PostGIS Part 1

git pull

Type Modifiers


data type

subtype type modifier

SRID type modifier

Geometry: Points and Linestrings




  • A point in 2D space specified by its X and Y coordinates
  • A point in 3D space specified by its X, Y, and Z coordinates
  • A point in 2D space with a measured value specified by its spatial X and Y coordinates plus an M value
  • A point in 3D space with a measured value specified by its X, Y, and Z coordinates plus an M value
  • A linestring in 2D specified by two or more distinct POINTs
  • A linestring in 3D specified by two or more distinct POINTZs
  • A linestring in 2D specified by two or more distinct POINTMs
  • A linestring in 3D specified by two or more distinct POINTZMs

Geometry: Polygons

  • Closed linestrings are the building block of polygons
  • A polygon contains all the enclosed area, and its boundary is the linestring that forms it
  • The enclosed linestring outlining the boundary of the polygon is called the ring of the polygon. More specifically, it's the exterior ring.

Collection of Geometries

  • A collection of geometries groups separate geometries that logically belong together:
    -  multipoints
    -  multilinestrings
    -  multipolygons
  • geometrycollection
    -  can contain any kind of geometry as long as all geometries in the set have the same spatial reference system and the same coordinate dimensions


  • In PostGIS, geography starts by assuming that all your data is based on a geodetic coordinate system, specifically the WGS 84 lon/lat SRID of 4326
  • Only include support for basic subtypes of points, linestrings, and polygons, and no support for anything above 2D space
  • The structure of the geography data mimic those of geometry
    -  everything you know about geometry applies to geography with no changes except for swapping out the term geometry for geography in both data type and function names


  • PostGIS raster data is stored in a table with a column of type raster
  • Data is usually evenly tiled so that one row holds the same rectangular size of pixels as other rows
    -  recommended to keep each row between 50 and 500 pixels for both width and height
    -  faster processing if large rasters were broken into tiles for storage in multiple rows rather than keeping them in a single row

Properties of Rasters

  • Raster width and height
    -  each raster tile (a row in the raster column) has a width and height that's measured in pixels
  • Bands
    -  each raster have multiple bands, but you must have at least one
  • Band pixel types
    -  rasters can only store numeric values in their pixels
    - pixel types: 1-bit Boolean (1BB), unsigned integer of 2, 8, 16, or 32 bits (2BUI,8BUI,16BUI,32BUI), signed integers of 8, 16, or 32 bits (8BSI,16BSI,32BSI), and two float types of 32 bits and 64 bits (32BF,64BF)

Properties of Rasters (cont'd)

  • Rasters and SRIDs
    -  georeferenced rasters have spaital coordinates defined within a SRS
  • Pixel width and height
    -  for georeferenced rasters, pixels do have heights and widths that reflect units of measure
  • Pixel scale
    -  if the width of each pixel represents 100 meters, we then have an X scale of 1:100
  • Skew X and Y
    -  the skew values are generally 0, and most rasters are aligned with the spatial reference coordinate axis, but on occasion they may be rotated


By Hui Hu


Slides for Lecture 3, Spring 2018, PHC6194 Spatial Epidemiology

  • 566
Loading comments...

More from Hui Hu