Ella Koeze

Lab 4 Questions

1. In your own words, explain the process of georectifying a digital image.

When georectifying a digital image, you first need to locate control points. We used Google Earth for this purpose, which roughly aligned the digital image with a base map. Using the transparency function, I identified road intersections or other easily identifiable points that existed on the image and the base map. I then found the geographic coordinates for those points on the base map.

Next, I added a basemap to a new map in ArcMap. I then added the digital image to Arc Map. I bookmarked the geographic coordinates of the predetermined control points I found in google map on the base map. I then used the control point tool on the georeferencing toolbar to connect the geographic coordinates with the intersection or other landmark on the digital image. ArcMap auto adjusted the digital image with each new control point, gradually aligning the map. Eventually, when I had about nine control points, I “rectified” the image on the basemap by using the “Rectifiy” command. This permanently warped the image to more closely match the basemap.

2. How might you or another researcher use this technique in the future?

If a researcher were gathering data through aerial images, for example, land use types, they might have to use this technique to georeference their data so it can be analyzed using a GIS.

3. Explain the different rectification methods available. Which did you choose?

You can rectify your image using an affine (first order polynomial), second order polynomial, or third order polynomial. They are used in various cases to more closely align the image with reality. The more complex distortion necessary, the higher order one should use, and the more control points one requires. I used affine because my image was not greatly distorted by topography and I only used nine control points.

4. What happened to the RME as you added more points? Why might this be?

As I added more points the RMS error increased and then stayed relatively stable. When I add my first few points, the image can is pulled more drastically to align with these points, increasing the error from the initial points. But as I add more points, the image aligns closer and closer to a particular alignment, and the error between all the points and the image stabilizes (unless I add an erroneous point).