5. Errors and Uncertainty

The occupancy type created some uncertainty by simplifying all the buildings in the database into only 6 classes.  This caused all buildings within an occupancy type to follow the same population distribution relationships for the three time scenarios.  Certain buildings are unique and do not follow the same relationships, for example, a church would have people only on Sunday morning and therefore does not follow the same relationship as most commercial buildings.  Stadiums are also unique because these buildings are dependent on the time of entertainment.  These types of buildings had to be classified as commercial for lack of a better classification.

Uncertainty exists in the estimate of population for the individual buildings.  Our building inventory was developed in 1995 and does not include all the buildings built afterwards.  As a result, our building inventory is missing many new buildings that have gone up in certain census tracts.  In order to determine building population, census variables had to be assigned to a smaller number of buildings, causing individual buildings to have higher values in these census tracts.  For example, certain census tracts contained very few commercial buildings but a large commercial population, causing the results to be overestimated. 

Additional uncertainty occurs when deriving individual building population based on the total area of a structure.  This allowed us to use the total area of a building as a ratio of the total area of all buildings, to distribute the census population evenly through our building inventory.  This method produces uncertainty because the size of a building may not accurately represent the number of people in that building at that time.  For example, an industrial building may occupy quite a large spatial area, while it is run by a relatively small number of employees.  In our analysis this building would represent a large portion of the industrial population of the census tract.

The census variables came from the Statistics Canada 2001 census data.  These results therefore estimate the number of injuries and casualties expected for the 2001 population.  Due to the steady increase in overall population, these values may underestimate current seismic risk.  By combining new building inventory data with present population data, the quality of the results would be improved, allowing the present conditions to be accurately modeled in GIS.