Enumeration areas are small spaces made up of one or more neighbouring blocks used for distributing questionnaires to households during census. Within urban areas, EA's do not necessarily reflect natural boundaries as they are designed to encompass generally a maximum of 650 dwellings. This results in a variety of differing shapes and areas.

Enumeration areas are dynamic in their design to changes in population and growth. Boundaries between census years are apt to change to meet changes in delineation criteria and are created using an automated delineation system. The overall goal is to create enumeration areas that are compact, accessible by road, respectful of standard geographic boundaries, and meet specified dwelling count and density criteria.

For the calculation of population densities, all but one of the above criteria compromise results. The result is a breakdown of space that has a variety of coverage. A quick look at an example of an enumeration area polygon superimposed upon a 1m geo-referenced aerial photograph (Image C1) illustrates this point.
Image C1
This image of the waterfront of New Westminster depicts an enumeration area that displays a very obvious variation in land use cover. The habitable space makes up only a fraction of the entire EA, the rest of the area is indeed the Fraser River. Of note is the density of actual living space; condominium complexes are distributed across the area in total.

The impact of this phenomenon is quite straightforward. A set number of people are divided by the enumeration area they fall within. This area includes space that is not and cannot be inhabited. The result is a density value that is understated and incorrect.

At the same time, enumeration area calculations of population density can be relatively accurate. This occurs in image C2 below.


Tract 012 exhibits continuity in land coverage. There appear to be equal sized houses built upon equal size lots throughout the area. Thus a calculation of population density is likely to be more accurate.

This differentiation of accuracy between enumeration areas introduces even more uncertainty for relating densities between two or more tracts. It is apparent that more precise land use data needs to be used for calculating population density.

The dasymetric process involves acquiring this information and incorporating it accordingly.

Dasymetric mapping is closely related to choroplethic mapping. The choroplethic method coincides data with data collection regions (for example the enumeration area) and thus closely reflects these divisions. When data collection units poorly match the character of a distribution, the choroplethic technique can hide and skew a good deal of information as we have seen above.

Using population density as example, researchers have noted three major problems with the choroplethic approach. Firstly, larger enumeration units tend to exhibit lower population densities and vice versa. Secondly, the enumeration units hide the variation that exists within them. Thirdly, enumeration unit boundaries are arbitrary, and are thus not likely to be associated with actual continuity in population density.

The dasymetric map offers an alternative to choroplethic maps by reducing this informational loss and ultimately improving the quality of both the analysis and product. The enhancing difference is the independence of mapping unit boundaries from enumeration areas. This is achieved through the incorporation of ancillary data that allows the analyst to create new, more natural homogenous areal units.

There are generally two strategies concerning ancillary data implementation used in dasymetry. The first incorporates, and is referred to as limiting variables, and involves an exclusionary factor. This form of ancillary data involves the inclusion of an outside source. This is the form demonstrated in this online example. The second utilizes and is known as related variables. This class of dasymetric technique is not demonstrated here. Using the original data only, related variables comprise a host of techniques and statistical processes to enhance an original coverage. These processes create the ancillary element necessary in dasymetry. This process is often associated with the processes of areal interpolation.

To conclude, a geographic information system can be used to increase the level of information for a spatial event. This project is designed to address this and to demonstrate a simple methodology for establishing better calculations of population density. The results of this method are then compared with the results of conventional population density calculations for evaluation over the same space.