Lab 3: Rectification in ArcMap

Overview:

The lab's goal was to to perform a geometric rectification of an old map of the Old Port region of Montreal. It had to be overlayed onto a current day map of the road network of Montreal. A network of control points were used to reduce the error margin of the geographical positioning of the map.

1.Georectify an old street map in Montreal

This process involves digitizing a rasterized paper map by implementing it into ArcGIS as a layer. Georectifying a map involves matching up raster data of the map with vector data in ArcGIS map.

1.1 Loading in the images

Both the shapefile dataset for the current streets of Montreal and the 1903 .png map were loaded into ArcMap. Since the 1903 image is not georeferenced, it was by default placed to coordinates of latitude and longitude equal to zero which happens to be near the equator south of England. Zooming to the layer extent makes it visible.

1.2. Labelling the current streets map

This was completed by changing the label field in the layer properties and selecting the "Label features in this layer". This process is seen in figure 1 below and the rendered map is seen in figure 2.

1.3 Rough estimate of old map positioning over the street network

Using the georeferencing toolbar, the "Fit To Display" command in the drop down menu transposed the 1903 map over the shapefile of the roads with approximately the same sizes to make it easy to work with. From this point, several other maneuvers using the rotating and shifting tools can be performed to slide or rotate the image to its approximated position. Figure 3 shows the effects of the "Fit To Display" command and figure 4 shows the estimated positioning of the map over the road network.

1.4 Using control points to align the maps

The location was of the map was further refined using the "Add Control Points" tool. This was done until 25 connected points were, as much as possible, evening scattered across the map in a roughly triangular pattern and the offset appeared to be a minimum for many points in a row. While using control points as the method to georectify an image, the object on the image must first be clicked followed by the same object in the shapefile. Figure 5 shows the map with all the control points. The points were placed in a roughly equally spread out triangularly manner to cover the most area as possible with the 25 points. The exercise was stopped at 25 points since the concept was learnt by then and the accuracy, as seen in figure 6, was sufficient for the purpose of the lab.

1.6 Using transformations for dealing with map imperfections

On certain locations of the map, its fibers were stretched and there were significant displacements in the data. Different transformations like 1st Order Polynomial affine, Adjust. and Spline, to name a few, were tested. The Spline transformation was chosen as the final one since it seemed to provide the best fit for the data. This sets the source control points exactly onto the target ones. Although accuracy at regions away from control points is not guaranteed, these regions were observed and considered okay for the sake of the lab. Figure 7 and 8 show the Spline transformed data to visualize its accuracy.

1.7 Rectifying the image

Rectify from the drop down menu was selected and a 5m cell size was selected. A new data frame was created and the rectified map was added to it. Figure 9 shows the rectified map.

2. Viewing the image in Google Earth

2.1 Steps for loading map to Google Earth

The "Layer to KML" tool was used to convert the map to a KML file with the .kmz extension. This is usable by Google Earth.

The map was then uploaded to Google Earth and viewed as shown in Figure 10.

Conclusion

The lab was successful in demonstrating the many tools that can be used to georectify an image on ArcMap. It allowed for a deep understanding of the Control Points too to be gained and demonstrated its usefulness.