Revised January 15, 2024

GEOG 3040/CV ENG 3050 – Intro to GIS

Lab 8 – Georeferencing and Digitizing

Many of the questions and requirements for this lab are embedded throughout the lab document.
They have been bolded and italicized to help you find them so you can address them as you work
through the lab assignment. You will also find a recap of the questions and requirements at the end
of the lab document.

Required Data:
Download the lab 8 data here: https://mailmissouri-
my.sharepoint.com/:u:/g/personal/blodgettc_umsystem_edu/EWJOUQGhJtFEuEHRqvZy19ABV
a-_G4dG_TPrx5XblIPmjw?e=VhRilK. The data is also available in Canvas within the Lecture 8
module.

Objectives & Background:

The objective of this lab is to teach you how to georeference image data that does not have a
projection, and then build your own datasets through digitization. Images are a common
source of information in GIS research and are particularly useful when mapping vegetation,
wetlands, and changes in land use. Although many images are already digitized and registered
to a projected coordinate system, not all are. This is particularly true for historical images,
which are often only available in physical, non-digitized form, and must be scanned to create a
digital version. Scanned images must then be georeferenced. Georeferencing an image, often
called an image transformation, converts the image from a file or scanner coordinate system to
a chosen or default projected map coordinate system.

There are many forms of image registration, but the simplest is called a first-order or affine
transformation. Affine transformations are best used when terrain is flat, and photos are taken
with a vertically oriented mapping camera (a Nadir view). When the camera is not pointed
straight down, then at least a projective transformation should be used, and if there is
significant surface height variation within the photo, then a complete geo-correction or photo
resection should be performed. In this lab we will just cover the affine transformation.

Image georeferencing requires both a source for geographic coordinates (our control layer) and
a target layer to be transformed (our image layer). Target layers may be images or
point/line/polygon feature classes. Our target is an aerial photograph in an arbitrary scanner
coordinate system.

The control layer (our roads vector layer) is the source for your control information. This
control layer must be in the desired map coordinate system. You will identify points in the
control layer and match them to corresponding points in the target image. When you have

1
Revised January 15, 2024

enough points and are satisfied with them and the subsequent error of the transformation, you
create a transformed image.

ArcGIS Pro:
Open ArcGIS Pro, choose a blank map and then add Broads_basic.shp. This is our control layer
from which we will take control locations. Be sure to add this data first. Next, add our target
layer Columbia1969.jpg. If a "No Coordinate System" error message appears, you can ignore it
because the point of georeferencing is to provide that coordinate system. If a dialog box asks if
you want to "Calculate statistics", choose No.

Right-click on each file and select Zoom to Layer to go to that layer's extent. What is the range
of coordinates on the roads layer? What is the range of coordinates on the initial image
layer? Next, use the Full Extent button to show the full extent of both layers simultaneously.
Notice how far apart the two data layers are in this view. If you can't see both layers at once–
or if they're too small to see–that shows how different their coordinates or locations are.
Right-click on the Broads_basic.shp layer and select Zoom to Layer. This will put the roads into
the center of your display–we need to see all the roads to begin Georeferencing.

In your Contents pane, make sure Columbia1969 is selected and click on Imagery
at the top, then click on Georeference to open its toolbar. See the figure to right.

Review the Georeferencing toolbar (shown in the figure below). In ArcGIS, use your mouse to
slowly hover over the toolbar buttons and discover what they are used for.

Click on the Fit to Display button. See the figure to right. This should bring our target
layer (Columbia1969.JPG) into the same extent of our control layer (our Broads layer).
The image should now appear with the Broads layer, but it needs to be rotated. See
the figure on the top of following page.

2
Revised January 15, 2024

In the Prepare portion of the toolbar, click the triangle to the right of the Fixed Rotate button
to open its dropdown menu and select Rotate Left. This will bring the target and control layers
closer to being coincident. See the figures below.

Next, we'll attempt to manually

rotate and move the image to fit
the roads. Use the Rotate cursor
to bring up the rotation tool. See
the figure on the top of the
following page. Try and move the
picture to match the roads.

3
Revised January 15, 2024

You will notice that you cannot really achieve what we want. Sometimes you get close to one
road, but at the expense of misalignment with others. Once you finish trying to rotate the
image to match, click the yellow dot in the middle to disable the rotate tool. Briefly explain
why you cannot get the image and road file to align properly using these tools in this manner.

OK, time to reset! The only way to clear your manual adjustments is to remove the
Columbia1969.jpg from your Contents pane and add it back again. You will receive a warning
when you go to remove the Columbia1969.jpg image. See the figure below to the left.

Click Yes to proceed, then add the image back into

your map, repeat the zoom to roads layer, select the
image layer, and in the imagery tab, click
georeferencing; and in the georeferencing tab, fit to
display, and rotate to get back to where we were at
the top of page 4.

Now we can begin the georeferencing process. To georeference our image, we

will start by using the Add Control Points tool. Click on the icon to activate the
cursor. See the figure to the right.

The cursor should now show a + and read from point (source). Find a location on the aerial
photograph that you can also find on the roads layer and click that location. The cursor now
changes to point (target). Now click on the corresponding point on the roads layer. You may

4
Revised January 15, 2024

need to change the Symbology of the roads layer to see it better in the viewer. Repeat this
process to create a succession of control links, or points, between the source layer and the
target layer.

Note that the selections should always proceed as follows:

• First, select a point in the image layer (the source),
• Second, select what you believe to be the corresponding point in the vector roads layer (the
target).

Zoom in on the Map to a portion of the image with a road intersection. Zoom in far enough
that you clearly see the width of the road, and easily position the cursor in the middle of the
road on the image. See the figure below.

Carefully place the cursor in the center of the road intersection in the image and click once.
Next, move the cursor over the corresponding road intersection in the vector layer (in this
example, the green polylines) and click again. If done precisely, you should see the image shift
or stretch to match the vector layer a bit better each time (sometimes it will only slightly shift,
other times more so).

5
Revised January 15, 2024

It is best to pick well distributed points, so for your second point select an intersection in a
different part of the image. Think about positioning these control points in a balanced fashion
across all four quadrants of the image.

Repeat this procedure, panning and zooming to a features that you can see or locate on BOTH
the image and roads layers. With each successive control link, the image should line up a bit
better.

If you would like to delete a connection point, click Select control point within the Review
portion of the ribbon and then click the point you would like to delete. You may now click
Delete to remove the control point. See the figure below to the left.

After you enter your 4th control point, examine the

control point table by clicking on the Control Point
Table button. See the figure to the right. Your
control point table should look like the one below.

Notice you get a report of the RMS error. This report provides an index of the accuracy of your
transformation. See the figure below.

Take a screen capture of the control point table with only your first 4 points and include this in
your lab document.

What is meant by the term "residual" in the table? What does it show? Be specific.

Strive to keep your RMSE number below 15, and ideally, try to keep it below 10. To achieve this
result, you will need to iterate the process of adding control points and then deleting the ones
that produce the greatest amount of error.

6
Revised January 15, 2024

Remember to keep your points well distributed and try to have three points in each quadrant of
the image. Again, this will be difficult to achieve because not all areas of the image have decent
control points available.

Take a screen capture of your final control point table with all your points and include this in
your lab write-up. What was your final RMSE value?

When you're happy with your control points, export your points table from within
the Georeferencing tab. See the figure to the right. Be careful to note the file
name and location, e.g., Col1969_Affine.txt. This will save a text file of your
control solution BUT it does not actually rectify the image to this control yet.

To import those saved control points in case you deleted some or would simply
like to pick back up on georeferencing you started previously, select Import
Control Points from the Georeferencing tab and then select the text document
that you previously created. See the figure to the left.

To create an adjusted
image, we must do the
following. First, from
the Georeferencing tab,
select Save as New. This
will bring up a dialog
box. Select your Output
Location (wherever you
prefer to store your
data), specify the TIFF
Image format, and give
the output a name such
as Columbia1969utm.tif.
See the figure to the
left. Under Settings for
the Export Raster,
ensure that the
Resample type is set to
Cubic. If we do not do
this our image might
contain artifacts. Click
Export to create the
new image. Click Close
Georeference to quit

7
Revised January 15, 2024

the georeferencing tool. You may be asked to save your work before exiting. This would
probably be a good idea. Zoom to the extent of the new image layer. See the figure below.

Assess how accurately your new image lines up

with the roads layer. Did you do a good job?
Are there any areas that are better from a
spatial standpoint than other? Explain. Be
specific.

We can now move on to digitizing the boundary

of an old quarry, some ponds, and an airstrip
from the rectified image. To accomplish this, we
need to first create a blank feature set using the
Catalog Pane.

From the Catalog Pane, right-click on your

working folder, select New, Shapefile. Name it
Polygons_1969.shp and import the projection
from the roads layer. See the figure to right.

If the new shapefile is not added to your map,

add it now.

8
Revised January 15, 2024

Select the empty polygon layer and open the Edit tab. From there, click on Create on
the Features subtab. See the figure to the right.

Note that the Construction Tools are included in

the create feature window. If you don’t see
them, double click on the filename and they will
open. Select the Polygon tool as seen in the
figure to the left.

Start digitizing with the pond(s) as your target.

Zoom in sufficiently so you can see the object you
are working on. Next, digitize the boundary of
the airport as well as the quarry(ies). Once you
finish each feature, double-click to connect the
first and last point which will close the polygon.

It is important to go to the Editor Toolbar and

Save Edits often. Once you have finished, then
click Stop Editing.

Take a screen capture or export a version of your final digitized polygon file (Ponds; Airport;
Quarry) with the image behind it and include it in this lab write-up.

What issues did you encounter while digitizing your three distinctive features?

What did you do to remedy the issues you outlined in the prior question?

9
Revised January 15, 2024

Here is an example of what one of the pond polygons may look like while building your
feature class.

10
Revised January 15, 2024

RECAP OF REQUIREMENTS & QUESTIONS FOR LAB 8:

Q: What is the range of coordinates on the roads layer? What is the range of coordinates
on the initial image layer?

Q: Briefly explain why you cannot get the image and road file to align properly using these
tools in this manner.

Take a screen capture the control point table with only your first 4 points and include this in
your lab write-up.

Q: What is meant by the term ‘residual’ in the table? What does it show?

Take a screen capture of your final control point table with all your points and include this in
your lab write-up.

Q: What was your final RMSE value?

Q: Assess how accurately your new image lines up with the roads layer. Did you do a good job?
Are there any areas that are better from a spatial standpoint than other? Explain. Be
specific.

Take a screen capture or export a version of your final digitized polygon file (Ponds; Airport;
Quarry) with the image behind it and include it in this lab write-up.

Q: What issues did you encounter while digitizing your three distinctive features? What did you
do to remedy the issues you outlined in the prior question?

11