Lesson 3
Georeferencing Raster Images
Section 1 - Screen Captures of the data frames after georeferencing the images:
Figure 1: State College DRG Data Frame
Figure 2: State College Map Data Frame
Figure 3: State College 1963 Photo Data Frame
Note: In this screen capture the map units still read "Unknown" because at the time of this screen capture the data frame had not been defined. Highlighting the data frame and then picking "Properties", the data frame can be defined as the NAD_1983_StatePlane_Pennsylvania_North_FIPS_3701, as are the image and shape files within the data frame. See Figure 4 below to verify this update.
Figure 4: State College Photo 1963 Updated Data Frame with Projection Information
Section 2 - RMS Error Calculation:
If the 'rule of thumb' for
georeferencing raster image files is that the RMS error be less than or
equal to one-half the side dimension, in map units, of a cell (pixel), then
my calculations are:
250 cells (pixels) per inch for a DRG
DRG Scale of 1:24,000
therefore - 96 inches per cell (24,000/250)
If 1 cell = 96"
then,
1 meter = 39.36 inches
1 cell = 2.439 meters (96 inches/39.36 inches)
1/2 cell = 1.2195 meters
or
acceptable RMS is less than or equal to 1.2195
Section 3 - RMS Error Discussion
In the three cases at hand in this lesson, we had copies of digitized maps or photos. In order to georeference these documents we need to be able to tell the GIS software what the map projection/coordinate system really is. If we do not know the projection of the original map or photo, it will be impossible to tell the GIS what is is. Further, even if we do know what the map projection is, but we do not have control points in the same coordinate system, we cannot tell the GIS what the projection/coordinate system needs to be.*
State College DRG - Both the map and the control points were given in UTM (meters) coordinate values resulting in a valid calculation for RMS. The achieved RMS value of .82467 is less than the calculated RMS limit of 1.2195 making this a valid calculation within the range of acceptable error.
State College TIF - The base map, though stated to be a USGS DRG, had both its georeferencing information removed and had obviously been physically modified as evidenced by the wrinkle in the upper portion of the map. Without any valid georeferencing information, the RMS error calculation is meaningless.
State College 1963 Photo - As stated in the text of the lesson, the photo has not been orthorectified to correct any scale variations that might be inherent to the aerial photo. Because of this, the photo itself is not georeferenced and as such fails to meet the criteria for a valid RMS error calculation - we cannot be sure of the accuracy of the photo.
Section 4 - What Might Limit An Ability to Arrive at a Low RMS When Georeferencing Raster Image Data
An inability to achieve a low RMS value when georeferencing raster image data can result from (1) incorrectly correlated and digitized ground control point locations in either the source or projected data due to operator error, (2) a poor choice of reference point locations, and (3) damaged or wrinkled hard copy source data. Poor eyesight, shaky hands, fatigue, lack of attention, misidentification of a control location, or just plain making a mistake when identifying the control points are the various forms of operator error that can contribute to poor RMS values. Additionally, the accuracy of the base map or the accuracy standards used when creating the original map can similarly contribute to poor RMS values. If you do not know the projection of the paper or scanned map, then you cannot be certain of the map units and cannot accurately calculate the length of the side of an image cell in order to arrive at a meaningful RMS value.
*Georeferencing Raster Images, Geography 484, Penn State University World Campus, https://www.e-education.psu.edu/courses/geog484/L03_cg.html, 27 October 2005.