Data Manipulation

Projection/Transformation of the Datasets

     The datasets we were given were not projected and needed to be given a projection. The dataset was taken from AutoCAD files and there was a problem with the correct georefencing of the datasets to the airphotos. We used the airphotos of Burnaby Mountain to verify that the datasets were correctly georeferenced. When we displayed the datasets and the airphotos together, we found that the datasets were about shifted approximately two kilometers away from the airphotos. To fix this, ArcInfo was used to translate the dataset from its current position to a new position which aligned correctly with the airphotos. ArcInfo has a command called TRANSFORM which changes coverage coordinates using an affine, similarity or projective transformation based on a set of control points (tics). We used the similarity option which would translate the coverage coordinates without skewing the data. The requirement for the TRANSFORM is that we need to create an output coverage with the corrected tic points. The TRANSFORM function will use these tic points as the parameters for the transformation equation which translates the dataset to the new location.

     The projection of the dataset was a simple task. Since the shapefiles were not projected (projection was specified as Geographic) we needed to project the data into UTM coordinates. The ArcInfo command PROJECT was used to project the data. Before projecting, the shapefiles were exported to ArcInfo coverages using ArcCatalog. We could only do the projection on dataset if it was in coverage format. We needed to create a projection file that would transform the input projection (Geographic) to the output projection (UTM). That projection file, geog2utm.prj, is shown below:

/**** geog2utm.prj *****/
Input
Projection Geographic
Units DD
Datum NAD83
Parameters
Output
Projection UTM
Units Meters
Datum NAD83
Zone 10
Parameters
End

Redigitizing of SFU Buildings and Other Layers

     We discovered that for some of the datasets it would be more efficient to redigitize the data over again. One of the datasets that had problems was the SFU buildings layer. This layer did not have separate polygons for many of the buildings and also were missing a few buildings. We also had to digitize a new layer called “fields” which combined all the playing fields on campus as well as the AQ pond. For the campus roads we decided to create two layers, one for the larger public roads and the other for the smaller restricted-access roads. We used ArcEdit for all the digitizing work. To redigitize the linework we needed to use the linework from the original AutoCAD dataset.

Creating the Static Viewshed Images in ArcView

     Besides the viewshed analysis using QTVR, we wanted to include static viewshed image for each viewpoint to show the visible surrounding areas. The static viewshed images give the user a map of the areas that are visible from the specific ground viewpoint. We only created viewshed images for the ground viewpoints and not the ones that were taken on the buildings since we do not know the exact elevation of these points. ArcView has an extension called 3D Analyst that allows a viewshed to be created from a Triangulated Irregular Network (TIN). ArcView allows a viewshed to be created from one point or many points. We only wanted to create a viewshed for a single ground viewpoint therefore we had to create shapefiles for each one. ArcEdit was used to extract the viewpoints from the master viewpoint coverageand put them into individual coverages. These coverages were then exported back into shapefiles to be used in ArcView.

     One preliminary step to do is to create a TIN surface from a layer with elevation points. We used the contour line shapefile to create the TIN surface. The TIN acts as the elevation surface from which the viewshed is determined. We now have a TIN surface as well as the shapefiles for each individual ground viewpoint. The viewshed image was created by simply selecting both the TIN surface and any one of the shapefiles (containing a single viewpoint). One thing we noticed was that viewshed created using the default one meter offset from ground level does not show a lot of visible areas. We tried using a two meter offset and this shows slightly more visible areas as expected. The two meter offset was favoured over the one meter offset.