This Section:
Data
Acquisition
Methodology
Data
requirements for this project at first seemed very innocent, yet the details
of collecting and manipulating the data sets proved to be very difficult.
It has truthfully been said that 90% of the time spent on a GIS project
consists of simply collecting the data and getting it into a useable form.
This proved to be the case.
There were some data sets that would have been useful, but could not be
acquired in time to perform the analysis. In some unavoidable cases surrogate
data was used, which may have degraded the overall result. Generally, however,
the data requirements for at least the basis of this project were met.
Following is a summary of the data I collected, their sources, and relevant
information regarding formats, and the like, wherever available. I have
attempted to trace the lineage of each of these datasets as far back as
possible, in order that we can trace the number of "generations"
of error we are dealing with.
| Data | Filename | Success? | |
| Elevation data
(eg. contours, TIN, DEM) |
2m_contour.dxf | The use of this .dxf file
as a source for elevation data presented a few
serious problems. Most importantly perhaps is the fact that the .dxf file unfortunately consisted of linework only, and it would therefore have been necessary to assign elevation values to every arc. This would have been possible, as the contour interval was known to be 2 metres, and the shoreline is reasonably accurate, however the sheer volume of this .dxf would make this a very time-consuming task. Also, the actual linework, while very accurate, is perhaps too accurate, as it includes in great detail elevation information that would actually hamper analysis - such as pits in construction sites that would later have to be removed. Nonetheless, this dataset is very valuable and was used as a secondary visual guide layer to aid in the digitizing of the streams. (see below) |
|
| gvrd_dem.rst | This was originally viewed as a
"fall-back" dataset, and was seen as being
of inferior quality to the 2 metre contour information, due to the poorer resolution This data was found on the S:/355Data/ general data drive, and is a gridded DEM of the Greater Vancouver Regional District at 25 metre resolution. (TRIM data) One problem with this dataset is that it features a "split" - or a groove of abnormally low values running north-south at approximately Crown Street. This derives from the fact that two TRIM datasets were combined here. (this split is shown in the following section) |
||
| note that several other datasets were derived from the above DEM. | |||
| Streams Data | DFO | I attempted to get streams data
in digital form from the Department of
Fisheries and Oceans, after receiving none from DFO directly, I instead decided to simply digitize the paper map that I had (below). |
|
| Digitizing | As a fall-back to using the DFO
data, I scanned the Vancouver section of the
"Lost Streams" map, and manipulated it using Adobe Photoshop 5.0 in order to sharpen it. I then brought it into ArcView GIS 3.2 (whose vector capabilities surpass those of Idrisi 32), where I digitized the streams and former coastlines of the City of Vancouver. Before digitizing, however, I registered the actual scan image (as if it had been an actual orthophoto), and - for good measure - included the now-registered contour data (above), which was used as a secondary guide, to be referred to when in doubt. This allowed me a greater accuracy in digitizing, as per the convergence of information principle. |
||
| Municipal Boundaries | city.rst | One source considered for municipal
boundaries was this raster image, found
on the S: network drive. It shows the municipalities of the GVRD as patches of raster cells. |
|
| muni.shp | One other source considered for
municipal boundaries was an ArcView .shp
which represented the municipalities of the GVRD as polygons. This was reprojected (in ArcView) from NAD27 to NAD83, and then imported to Idrisi using SHAPEIDR, where it was examined, and rasterized using POLYRAS, creating the MUNIRAS image. |
||
| Bikeways and Greenways | bikeway.shp | This file is an ArcView shapefile
created in 1999 for a previous GIS project
concerning the development of Vancouver's bikeway and greenway system. It consists of linework showing Vancouver's system of bikeways and greenways. It was reprojected in ArcView from NAD27 to NAD83, and imported using SHAPEIDR. In Idrisi, it was examined, and rasterized using LINERAS to create the BIKEWAY image. |
|
| Soils and Geology | soils.rst | This dataset consists of soil classifications
for the GVRD, which are shown as
patches of like raster cells. One problem with this is that cell values are discrete, and therefore - given the continuous nature of soils data - we are confronted with a very large number (nearly 50) of categories with very similar names such as "silty clay", "clayey silt", "silty clay to silt", and "clayey silt to clay". This dataset was useful but unwieldy and - at first - difficult to manage. No importing or reprojection was necessary. |
|
| Land Use / Zoning | landuse.rst | This dataset consists of land use
classifications throughout the GVRD. It is
already in Idrisi format, so no importation or reprojection was neccesary. |
|
| zoning.TAB | This is a MapInfo file showing
zoning areas in Vancouver as polygons. It was
originally intended to complement the landuse.rst data above. It was reprojected and converted to a .mif file in MapInfo. An attempt was made to import this data into Idrisi using the MIFIDRIS module, but this invariably ended with a crash. Other attempts were made to bring the data "through" Arcview GIS and use SHAPEIDR to import, but this netted the same unfortunate result. Zoning information was unfortunately excluded, and landuse was relied on as a surrogate. |
||
| Block Outlines | block_outlines.TAB | This is a MapInfo file (originally
a .dxf) showing, in great detail, the city blocks,
lot lines, alleys, and srteet rights-of-way for the City of Vancouver. It would have been extremely valuable in this analysis, however, repeated attempts to import it into Idrisi were similarly met with repeated crashes. Attempts to import this data "through" ArcView GIS appeared to be initially successful, but again resulted in a crash. Unfortunately, this data could not be included in this analysis. |
|
| Bridges | bridges.TAB | This was grouped with the block_outlines.TAB,
and met similar results. It would
have been used as a reference point for maps and display purposes. |
|
| Parks | parks.shp | Although parks were included in
the landuse.rst file, they were not included at
a sufficient resolution for this analysis. When dealing with urban stream restoration, it is often the smaller, locally-important parks that can be the most useful as sites for urban stream daylighting. The parks.shp dataset was compiled for a previous GIS project, and I am reasonably certain - having compiled it myself - that it includes every park in the City of Vancouver. This data was reprojected from NAD27 to NAD83 in ArcView, and imported to Idrisi using SHAPEIDR. |
|
| Rainfall | (none) | No precipitation data ws acquired
in good time for analysis. In place of this, aspect was used as a surrogate,
on the understanding that, generally in Vancouver, south-facing slopes
get more rainfall than north-facing ones. Aspect was used as a surrogate,
but
a lower weight was given. |
|
The basic methodology of this analysis relies on a series of weighted multi-criteria-evaluations
and overlays based on fuzzy sets. This allows the MCE to take into account
a wider range of possibilities than would be possible with a strictly boolean
analysis.
A cartographic model has been produced which clearly shows these steps.
There is also a textual summary which follows.
A successful urban stream restoration project depends on numerous factors, which might include:
1. a high slope, both to ensure that the stream does not stagnate,
and because areas with high slope often have the
worst problems in dealing with stormwater runoff and the erosion caused.
A high-slope area would also likely be easier
from an engineering standpoint, as the "daylighting" would neccesarily
involve the digging up of a previously underground
waterway.
2. an area with high rainfall, as these areas are often the ones in which stormwater management is most crucial.
3. an area with suitable soils, to ensure that the stream is able
to fulfill its hydrological role, and that the actual digging
operations will be inexpensive. Different soil types also reflect on an
area's ability to support riparian vegetation.
4. inclusion in a park, which would make the entire daylighting process much easier from a legal (land tenure) view.
5. suitable land use characteristics, which would ensure that the
stream would indeed become a focal point for a local
neighbourhood, so as to increase exposure and promote urban ecological
awareness. (so, for example, industrial
areas would not be considered as highly as residential areas.)
6. proximity to an existing waterway or shoreline, which would make
sense from a holistic point of view; if a stream is to
be daylighted, it should ideally be connected to another existing waterway
or shoreline.
7. proximity to a bikeway or greenway, which would provide for greater
public access on foot or bicycle. It may also
provide an alternate route for an existing greenway.
8. distance from a bikeway or greenway,
which may (despite the above) also be valuable, as this could provide a
right-of-way for a new greenway in an area previously unserved by the greenway
network.
9. distance from a major roadway or highway, which would ensure
that the linear park created would be accessible to
pedestrians and cyclists. This would also reduce the impact of noise and
exhaust.
These factors were operationalized through boolean and fuzzy set logic,
and were eventually combined using a series of weighted MCE's, overlays,
and orthographic displays.
The orthographic displays are included not so much for the sake of analysis,
but as a powerful visualization tool. 2.5-D surfaces are very good at communicating
data, and serve to illustrate very well the reasons for pursuing a particular
project. Such displays are not ends in themselves, but rather they are
tools in visualization and education. A good illustration would be the
ortho images of Vancouver showing the locations of the lost streams. This
type of visualization was identified as a goal for this analysis, in order
that the image might be used to promote awareness of urban ecological issues.
The exact methodology is outlined in detail (by a cartographic model) in the Data Manipulation section.