Chapter 5 - Georeferencing

Georeferencing is a broad but very important area of GIScience that allows us to associate features with reliable spatial data for mapping, queries, and analysis. Finding or adding spatial data can sometimes serve as a very useful way to organize disparate information.

We discuss this chapter now because it adds depth to the GTKAGIS chapter on Projections. The formal presentation in Section 5.7 should be read in conjunction with GTKAGIS because each has useful visualizations.

5.1 Introduction

The chapter begins with a reference to "location, time, and attribute" but I prefer to think of the 4 dimensions of SPACE/TIME/PHENOMENON + SCALE. You needn't worry about the details of this triangle except to keep the idea of of WHERE/WHEN/WHAT clear and separate in your mind, especially when examining data. And although time may be regarded as optional, all data were collected at some time and because phenomena change, time is often critical.

To understand the value of FIPS codes see how many records you get if you run the following selection:
    Select * FROM USA.counties WHERE: "NAME" = 'Washington'
versus
    Select * FROM USA.counties WHERE: "FIPS" = '01129'
Both are nominal variables, but only one is unambiguous (and cannot be misspelled!).

Box 5.1 may seem familiar because we already looked at an even more elegant multiscale system in Section 10.7.2.

Table 5.1 may make a bit more sense if you associate the rows with the following geometries or data types:

      SYSTEM GEOMETRY   DATA TYPE
1. Placename
2. Address
3. Postal code (ZIP)
4. Phone area
5. Cadastral
6. PLSS
7. Lon/lat
8. UTM
9. State Plane		 point(s)
line
polygon
polygon
polygon
quadrangle
point
point
point

vector
vector
vector
vector
(lon, lat)
(EAST, NORTH)
(EAST, NORTH)

5.3 Postal addresses and postal codes

GTKAGIS Chapter 17 is an exercise on geocoding that illustrates that an address yields a coordinate based on an interpolated location on a line segment of a street polyline. We do this in order to give an address a physical location in space for further analysis. The result should be close to what you'd get if you took a GPS receiver to the door.

Postal codes are strings like FIPS codes. Although they are not integers, if they are treated as such by a database they cannot be joined to data. FIPS codes have the added disadvantage that some begin with zero, so if they get interpreted as integers the first character is dropped so that  Jefferson County Alabama, whose FIPS code is '01101' becomes the integer 1101. I have spent hours cleaning databases after a program dropped the leading zero! (How might a really bloody-minded computer might turn this string into the integer 13?)

5.4 Linear referencing systems

I'm not sure why there is no geocoding example in this section, because Box 5.3 gives a verbal description of the process. Substitute "Broadway" for "Birch" in Figure 5.5 and you get the idea.

5.5 Cadasters and the US Public Land Survey System

The Township and Range system is yet another example of nested quadtrees we saw in Section 10.7, but less rigorously mathematical, and not extended to infinitely finer resolution. If you travel by air across the Western U.S. you'll see these "sections" passing your window about every 6 seconds (why?).

5.6 Measuring the Earth: latitude and longitude

The first two figures in GTKAGIS Chapter 13 are a clearer illustration of how longitude and latitude are simply angles. A little-known but quite useful fact is that the meter was originally defined as 1/10,000,000 of the distance between the pole and the equator. So one degree corresponds to
    10,000,000 meters / 90° = 111,111 meters/degree
but note that this is only true of degrees of latitude and not longitude (except at the equator). This is yet another argument in favor of the metric system. (I haven't read any of the source documentation on this, but the link above sounds authoritative.)

While we're at it, it helps to know that a nautical mile is 1 minute of latitude
    nautical mile = 111,111 meters / 60 = 1852 meters = 6076 feet
or about 2000 yards, so if you spend time moving on land and water in and out of the metric world you need to be able to do some quick math.

The rest of this section is a rather technical review of spherical angles; these matters are extremely important in such fields as cartography, surveying, and navigation, but when you have GIS projection problems they probably won't be solved with mathematical computations but rather by reading metadata, knowing the available projections - or knowing someone to email if you can't overlay your data.

Examine Figure 5.7 and figure out which sense (clockwise or counter-) that the earth would rotate? (And note that this is rotation and not revolution, which is what Earth does around the Sun.)

5.7 Projections and coordinates

Note the Greek letters used in these formulas; it is unfortunate that lambda corresponds to longitude and not the more mnemonic latitude, whose letter is phi.

Figure 5.12 is a more elegant presentation of the three major genera of projections than what's in the introduction to GTKAGIS Chapter 13, but the most powerful way to understand the effect of projections is to experiment, as we have done.

Perhaps one of the first explicit advances in GIScience (and which precedes digital GIS) is the idea that scientific measurement of absolute location, relative position, and distance required reliable transformations between angles and coordinates. This is why you should be aware that under the GIS hood some pretty complicated mathematics (and very fast computation) is going on.

Task A more interesting variation of GTKAGIS Chapter 13 would be a 2x2 layout illustrating what happens to the conterminous US (or any other large region) under 4 different projections. It's been done many times before, but it will certainly familiarize you with projections.

A particularly elegant projection system is Universal Transverse Mercator, which you can envision as resulting from wrapping a cylinder around the earth at a line of longitude (meridian): see Figure 5.12A but imagine the cylinder placed horizontally, so you get the zone shown in Figure 5.16 and the world map shown in Figure 5.15, which is a useful reference if you need to look up UTM zones.

You may encounter state plane coordinates but this is just another projection, albeit very idiosyncratic; Kristen Kurland and Wilpen L Gorr's excellent ArcGIS tutorials have Pennsylvania SPC data because that's where they work.

5.8 Measuring latitude, longitude, and elevation: GPS

Geographers have been making a big fuss over GIS, hoping among other things that it would give their field a big boost, but the real leap came with Global Positioning System (GPS) which put computer maps in and on millions of pockets and dashboards, closing the often mysterious loop between position, map, and coordinate.

On Saturday 2008 October 25 I led a George Mason University geography field trip in which we used a GPS receiver to mark the boundary stones of the original District of Columbia, You can copy these numbers into a spreadsheet and plot them or - more elegantly - read them into ArcGIS and map them (but the altitudes are not reliable). The point is that these coordinates are unambiguous, reliable, and global (except for issues of datum). If you don't have a handheld GPS receiver, I urge you to get one: not only will you find it useful, it will teach you a lot of geography!

5.9 Converting georeferences

Geocoding is an extremely useful GIS technique that maps addresses into coordinates, is fairly easy to understand as just another transformation, from lines to points, but it's devilishly tricky to implement (as you may have noticed from GTKAGIS Chapter 17) and somewhat error-prone. I'm not sure why this book doesn't present a simple example of the technique.