Welcome to Adobe GoLive 5

METHODOLOGY (pre-analysis)

1. Data Acquisition

I acquired a database file from John Perrie, the webmaster for the Vancouver Fire & Rescue Website, containing information about all incident responses by VF & R for 2001. John was both quite willing to help and knowledgeable about the information would be useful for my project.

In terms of spatial data, I used an ArcView 1996 Census Tract (CT) shapefile, and both Landuse and DEM layers in Idrisi format.

This software and data is made available to SFU Geography students. For more information, email Jasper, the SFU SIS Lab administrator.

2. Creating the 'Firescore' Layer

There were 1038 incident responses in Vancouver for 2001. Before I could import these into a spatial database, I needed to create several fields, leading to a final field firescore which would rank each CT based on its number of incidents, each of which would be ranked according to type and time from alarm to extinguishment.

The database contained the following fields:

INC_NO: A six digit ID number

EXP_NO: An incident with only one database entry was marked 0, and any subsequent incident related to this primary incident would be given a larger value. Only a few incidents were related, with the largest agglomeration having a total of 8 related incidents.

RPTD_DT: Reported Date

INC_TYP:Incident Type= reclassed in INC_T SCORE /10

00 (unclassified incident situation)= 3
01 (unclassified fire/explosion)= 6
02 (structure fire)= 9
03 (mobile property fire: structure)= 6*
04 (mobile property:inside a structure)= 6
05 (mobile property:outside a structure)= 5
06 (fire in trees,brush,grass or standing crops)= 6**
07 (refuse fire including rubbish containers)= 3
08 (other outside fire)= 5
09 (structure fire and Hazardous-Material incident)= 10***
10 (mobile property fire and Haz-Mat incident)= 7
11 (no alarm; no initial response by VF&RS)= 3

[Notes from John:

*often involves cars in garages or high rise garages, and usually involves almost as many hazards as regular structure fires

**"In our urban environment, these are usually fires in trees,
ornamental shrubs, hedges etc. that simply have a dollar value to the owner."

***"Every structure fire involves hazardous materials! These incidents are only ones where a known, extraordinary problem exists, like the involvement of dangerous chemicals, etc."]

LOCATION: Address of fire location

CENSUS: CT value. CT's are 6-digit numbers, with Vancouver's ranging from 933001 to 933068, with larger ones divided into 2 or more subdivisions. This would be the value field that would be needed to merge the database with Idrisi.

ATE: Time alarm-to-extinguishment= reclassed in ATE SCORE /10

0 (undetermined)= 2
1 (0 min.)= 1
2 (<2 min.)= 2
3 (2 - 5min.)= 2
4 (6 - 9 min.)= 2
5 (10 - 19 min.)= 3
6 (20 - 39 min.)= 5
7 (40 - 60 min.)= 8
8 (> 1 hour)= 10

TOTAL: Total score/20 for each incident

CT TOTAL: Total score for each CT by adding all its incident totals

3. Importing Census Tract Shapefiles from ArcView to Idrisi

Idrisi was to be the software of choice for project GUI output because of the project's emphasis on, and Idrisi's superior analytical capabilities of continuous surfaces. Idrisi's legends are usually created to show continuous real numbers, and while the data for firescore (the eventual output name for the value of CT TOTAL) is represented by whole numbers, this model is somewhat fitting (perhaps indirectly) as each CT score is made up of many weighted incidents, rather than simply a summation of whole numbers.

I chose to take Census shapefiles from ArcView GIS and Import them using SHAPEIDR.

Using Database Workshop, I edited the attribute table for the resulting layer to add the field firescore and enter the results from CT TOTAL.

Here was the resulting map:

*Note that fire data for the UBC grounds was not included in the map, since it is not considered part of the city of Vancouver and its CT was given the 'other' score of 0.

However, in reality, its score was a substantial 122.

CT 59.01, the CT containing the Downtown Eastside and part of the downtown core, skews the distribution somewhat. By altering the min-max settings, however, we can get a more balanced distribution.

Now begins spatial analysis. Click here, or navigate to any of the pages below.