Embedded Files
Lab Overview
Lab Overview
Objective: To use ArcGIS to manually digitize and georeference an image using ArcGIS rectification tools.
In Lab 2, ArcGIS rectification tools are used to transform a .jpg image of an 1903 Montreal street map into an ArcGIS layer, effectively merging the 1903 map with a vector shape file (.shp) of the Montreal street layouts. The result is a digitized version of the 1903 map with the same coordinate system as the vector street data. The resulting map is converted to a KML file for viewing on Google Earth.
1.0 The data
1.0 The data
In order to rectify the 1903 Montreal Map, we need both an image file (.png, .tiff, .jpg, or another image format) and a vector data .shp file to "anchor" the image file, allowing us to associate the image file with an established geographical reference.
1.1 Vector Data
1.1 Vector Data
We are provided with a .shp file that includes vector data for the streets of Montreal, projected in MTM zone 1. The units are in meters - it is important to note the units of the files used to prevent errors in analysis later in the project. Figure 1 shows a display of the vector street data.
1.2 Raster Data
1.2 Raster Data
The 1903 Map of Montreal is downloaded as a .jpg file. Figure 2 shows the file in its original orientation.
Tip 1: Order matters!
When adding data files to ArcMaP, the first data file added to ArcMAP determines the coordinate and projection system of the project. In this case, it was important that we add the vector .shp file first, as it is the only file that has an established coordinate system at this point in the project.
When adding control points, the .jpg image file should be selected first, the vector street image second, to avoid distortion of the results.
When adding data files to ArcMaP, the first data file added to ArcMAP determines the coordinate and projection system of the project. In this case, it was important that we add the vector .shp file first, as it is the only file that has an established coordinate system at this point in the project.
When adding control points, the .jpg image file should be selected first, the vector street image second, to avoid distortion of the results.
Figure 1: Montreal Vector Street Data
The above .shp file was used to rectify the 1903 image of Montreal presented in Figure 2
Figure 2: 1903 Montreal Map
Figure 2 is a .jpg file created from scanning a physical map of 1903 Montreal.
2.0 Rectification with ArcMap
2.0 Rectification with ArcMap
This next section details the rectification process and includes information on tips and lessons learned during the exercise.
2.1 Georeferencing tools
2.1 Georeferencing tools
Next, the vector and raster data are added to ArcMap. The 1903 street map is automatically positioned by ArcMap with it's bottom left corner at (0,0) coordinate - this happens to be nowhere near the projection of the current Montreal street data!
2.1.1 Rough alignment tools
The Georeferencing tool is used to shift and orient the .jpg image manually sot hat it roughly matches the vector data. The "Fit to Display" command transposes the image on top of the .shp file, then the rotation and move tool allow for more minor adjustments. Toggling the transparency of the image file helps with this orientation process.
2.1.2 More precise alignment tools
To refine the orientation process, the "Control Point" tool is used. The control points anchor the image to the vector map. Easily identifiable landmarks, like intersections of major roads, parks, are selected, first on the 1903 image, then on the vector datafile. The top image is "pulled" to the correct spot on the vector coverage. This is known as "rubbersheeting", as the image is stretched like a sheet of rubber over the base image, while the control points act like anchors, securing the image in place at manually determined points. Over 50 control points were added to ready the image for transformation.
2.1.3 Transformation tools
After adding the control points, the transformation tools were explored to make final adjustments to the 1903 image. First, Second, and Third Order Polynomial Transformations were used, giving the results shown in Figures 3, 4, and 5. These figures show that, for this particular map and distribution of control points, the First and Second Order transformations are more suitable. The Third Order transformation is well fitted in the SE corner (due to the concentration of control points here), but very distorted in the NW corner, due to a lack of control points in this area. This phenomenon speaks to the importance of ensuring an even distribution of control points, especially around areas of a map is especially important - a building site, for example.
Figure 3: First Order Polynomial Transformation
Figure 4: Second Order Polynomial Transformation
Figure 5: Third Order Polynmomial Transformation
Tip 2: Control Point Clustering
It is important to attempt to evenly distribute the control points. If they are clustered, as in Figures 3-5, when transformation algorithms are applied, the areas with fewer control points will be stretched and distorted, and the resulting rectification will be skewed.
It is important to attempt to evenly distribute the control points. If they are clustered, as in Figures 3-5, when transformation algorithms are applied, the areas with fewer control points will be stretched and distorted, and the resulting rectification will be skewed.
2.2 Accuracy evaluation
2.2 Accuracy evaluation
Due to wear over time, folds are visible in the image of the 1903 map, where the map fibers stretched and lost their pigment over time. This "stretching", though it is minuscule, increases the risk of misalignment. Examining this area provides us with a good estimation of how accurate the fit of our image is to our vector data file. Figures 6, 7, and 8 show the difference in fit between the three polynomial transformations in the area of the fold distortion. By observing how closely the the vector data (blue lines) follow the streets of the 1903 map, we can see that the First Order Transformation outperforms the Second and Third in this region of the map.
Figure 6: First Order Transform - Fold Distortion
Figure 7: Second Order Transform - Fold Distortion
Figure 8: Third Order Transform - Fold Distortion
2.3 Rectification
2.3 Rectification
Once the map has been georeferenced to our satisfaction, the "Rectify" tool can be chosen from the Georeferencing drop-down menu. The map was saved in GRID format (the ArcGIS raster format), as an 8-bit image (NoData value = 255). The result of the final rectification are presented in Figure 9 below.
Figure 9: Final Rectified 1903 Montreal Map
3.0 Accessing the map with Google Earth
3.0 Accessing the map with Google Earth
This step converts the new raster file to a KML file which is accessible through Google Earth, allowing for further sharing and analysis options.
3.1 Map to KML
3.1 Map to KML
Using the "Map to KML" conversion tool (in the Conversion toolbox), the map document is converted to a file type accessible through Google Earth. Now, the map can be added as a layer and accessed through the Google Earth platform. Figure 10 shows the map as viewed through Google Earth.
Figure 10: Rectified 1903 Montreal Map on the Google Earth Platform
References:
Cardille, Jeffrey. 2019. ENVB 530: Geometric rectification in ArcGIS 10. Retrieved from: https://docs.google.com/document/d/1wQyVz6GkfwDfnIg25Ijir-AHFbBJFMNADivycDGMfTY/edit
Google Earth Help. 2019. Import your KML map data into Google Earth. Retrieved from: https://support.google.com/earth/answer/7365595?co=GENIE.Platform%3DDesktop&hl=en
Done for Advanced GIS for Natural Resource Management, in the McGill University Department of Natural Resource Sciences, Professor Jeffrey Cardille
Google Sites
Report abuse