GPS Data Collection and Image Georeferencing

            Georeferencing is an important tool that involves applying real world coordinates onto a map, air photo or digital imagery. Raster data is commonly obtained by scanning maps or collecting air photos and satellite imagery. Scanned map datasets lack spatial reference information. Locational information for air photos and satellite imagery is inadequate with data aligning incorrectly with other data in GIS. Georeferencing allows the aligning of raster datasets in conjunction with other spatial data to a map coordinate system. The raster data can then be viewed, queried, and analyzed with other geographic data in GIS.

            One method in obtaining real world coordinates is measuring the coordinates based on satellite signals through GPS technology. The GPS system has been designed to provide information as accurate as possible. However, there are still errors that cause deviations from the actual GPS receiver position. Aside from practicing the importance of georeferencing, the purpose of this exercise was to evaluate uncertainties in order to minimize their effects and allow for improved accuracy.

Signal clarity and reception is not entirely accurate when utilizing GPS. For example, using GPS indoors can cause weak signals. The signal hits and is reflected off objects such as buildings and rocks. The GPS device can fail to give proper coordinates when situated in a low reception area. The built in clock of the GPS can cause errors as well when it is not as accurate as the atomic clock of the satellites. This timing error produces errors in calculations. Users of GPS can also generate errors. GPS users can make mistakes recording the exact location of where the coordinates were taken. Identifying a series of ground control points (GCP) links locations on a raster with that of the existing spatial referenced data. These control points can be accurately identified on raster datasets and real world coordinates. The GCPs ultimately minimize the errors associated with human induced mistakes. There will also be errors when matching the locations of the GCPs with GPS coordinates. These errors occur because it is rare and unlikely for the exact location on the map to be located. GCP distribution is important in the accuracy of the geometric correction process. It depends on the number of selected GCPs. The GCPs should be evenly distributed and have enough surrounding a certain section. For instance, section 1 of the UCLA campus can have 2 GCPs on the top and none at the bottom. Section 2, which is right underneath Section 1, can have 2 GCPs on top to make up for the missing GCPs on the bottom of Section 1. A bad location and distribution of GCPs cause an increase in the average RMS error value of correction.

            This lab assignment involved groups of four or five going out into the field and collecting points of interest throughout campus. Errors could have been caused by some groups not recording the correct coordinate information. Out of all these points collected, 16 points were chosen for georeferencing. A separate excel sheet was created with only these 16 points and then added as x and y data on ArcMap and defined as UTM Zone 11. The points were exported in a shapefile and fitted to display with the UCLA JPEG image. After completing the georeferencing process for all points, residuals and total RMS error was saved. Lastly, the image was rectified as a TIFF and the points were shown on the map as symbols representing residuals.  The street layer was also added and clipped with the rectified image.

            Results indicate residuals that show the difference between the original or actual point specified and where the point ended up. It also generated a RMS (root, mean, square) error, which is the total error computed by taking the RMS sum of all residuals. The RMS error indicates the consistency of the transformation between the different control points. Point 6 was removed to reduce the RMS error. However, keep in mind that a poorly entered control point can be the cause of drastic errors even when the RMS error appears to be low. The RMS error recorded, as shown on the map, was 4.77714. The RMS value is high assuming that the groups made mistakes and GPS also had inaccuracies. If there was more time to collect GCPs, the result of RMS could have been lower by allowing each group work with their own GCPs across the campus. Georeferencing is a valuable tool nonetheless. GPS technology should be constantly upgraded and improved to achieve greater accuracies, provided that there are ways to reduce errors.