Methodology

View the most recent version.

Overview
Ecumene creation

This section highlights the methodology for creating the 2006 population ecumene. Before providing the details, however, it is pertinent to relate the current methodology (which is very similar to the 2001 approach) to the geographic and cartographic issues described in Sections 2 and 3.

Three points stand out. Firstly, the ecumene concept is certainly not new, and the Geography Division is continuing a long-standing geographic tradition. Secondly, the Canadian ecumene delineated by Gajda (1960) and the settled area of Alberta defined by Kariel (1970) are quite similar to Statistics Canada's population ecumene. However, since the initial delineation of the population ecumene over 30 years ago for the 1976 Census, the vast expansion and improvements of our spatial digital database, the increased use of GIS software and the introduction of the block program for the 2001 Census all contribute to a more accurate and detailed ecumene. Thirdly, because the choropleth map has limitations, the use of a population or agricultural ecumene limits the statistical data to only where population or agriculture are found — resulting in a more accurate depiction of spatial distributions.

Overview

Spatial and attribute data from the Geography Division's Spatial Data Infrastructure (SDI)¹ is used to generate the population ecumene, namely dissemination blocks² as well as population counts, population density and land area of the dissemination blocks.

The population ecumene is constructed by selecting those dissemination blocks having a minimum population density of 0.4 persons per square kilometre (about one person per square mile). It is the small spatial extent of these blocks that allows an extremely detailed ecumene to be generated. The detailed ecumene limits are generalized to ensure visibility for small-scale thematic mapping. Where appropriate, edges are smoothed, small, neighbouring ecumene pockets are aggregated to form larger areas, small, isolated ecumene pockets are exaggerated in size, and extremely minute pockets are eliminated. There is at least one ecumene pocket (polygon) in every census division.

The Population Ecumene Census Division Cartographic Boundary File, Reference Guide includes details on the product as well as technical information for users (Statistics Canada 2007a). While Appendix 1 contains a technical description of the processing steps, the following section on generating the ecumene provides a more detailed explanation of the generalization procedures.

Ecumene creation

Using the dissemination block boundaries, every block is classified as being either an ecumene block (population density of at least 0.4 persons per square kilometre) or a non-ecumene block (population density less than 0.4 persons per square kilometre). This results in a population ecumene that is far too detailed for small-scale thematic mapping (Figure 4.1).

A series of generalization steps are necessary to render the ecumene more suitable for small scales. Contiguous blocks having the same classification are merged to create larger polygons. Three layers are created from this master file: the core ecumene, ecumene pockets outside the core ecumene and non-ecumene pockets inside the core ecumene.

The layer for the core ecumene is generated by selecting the merged dissemination blocks that are equal to or greater than 1,000 square kilometres. The size criterion assists in recognizing and detecting the large, core ecumene. Neighbouring polygons are further aggregated using the AREAAGGREGATE command. Part of the aggregation step involves converting the layer from vector to grid format according to a specified cell size and merges polygons based on a given distance between polygons. In all cases, this distance must be greater than the cell size³. The polygons are assigned a minimum cell size of 4,000 square metres and a distance between cells is set at 4,001 metres. Polygons less than 4,000 square metres are eliminated, resulting in a less complex pattern.

Unnecessary detail is removed using line smoothing (Environmental Systems Research Institute 1996, p. 4). The BENDSIMPLIFY option retains the essence of the original lines as opposed to the POINTREMOVE option. BENDSIMPLIFY reduces a line by detecting and removing extraneous bends from the original line, therefore preserving the main shape of the feature and cartographic quality (Environmental Systems Research Institute 2000, p. 7). The blocks are then 'cleaned' to remove any resulting slivers, buffered to 7 kilometres and the outline smoothed. Testing reveals that 7 kilometres produces the optimal generalization effect.

The layer for the ecumene pockets outside the core ecumene is generated by selecting the merged dissemination blocks that are less than 1,000 square kilometres. The size criterion assists in recognizing and detecting the ecumene pockets outside the core ecumene. The pockets are enlarged using a buffer of 5 kilometres⁴. The aggregation procedure merges groups of neighbouring pockets (Environmental Systems Research Institute 1996, p. 4) to create larger, more visible ecumene pockets that would be suitable for small-scale thematic mapping. The aggregation step is similar to that used for the core ecumene, by selecting a minimum cell size of 4,000 square metres and a distance between cells of 4,001 metres. Polygons less than 4,000 square metres are eliminated. As a result some census divisions no longer have an ecumene pocket; in these cases the pockets are reinstated. The grid layer is then converted back into a vector file.

The layer for the non-ecumene pockets inside the core ecumene is generated by selecting the merged pockets that are equal to or greater than 2,000 square kilometres. The size criterion assists in recognizing and detecting the non-ecumene pockets inside the core ecumene. The aggregation procedure is similar to that used for the core ecumene and the ecumene pockets outside the core ecumene. In this case a minimum cell size of 2,000 square metres is assigned and the distance between cells is set at 2,001 metres. The non-ecumene pockets are then buffered to 2 kilometres and smoothed.

Additional non-ecumene pockets inside the core ecumene are created when the 7 kilometre buffer is applied to the core ecumene, thereby closing off any gaps.

The three layers are merged (Figure 4.2) and the resulting single layer is visually checked.  Finally, an inspection is undertaken to ensure that at least one ecumene pocket exists in every census division. The resulting output is clipped to a generalized shoreline to produce the population ecumene cartographic boundary file.

---

Notes

1. The Spatial Data Infrastructure (SDI) is an internal database containing spatial and attribute data that support a wide range of census operations. It is also the source for generating geography products for the 2006 Census (Statistics Canada 2007b).
2. A dissemination block (DB) is an area bounded on all sides by roads and/or boundaries of standard geographic areas. The dissemination block is the smallest geographic area for which population and dwelling counts are disseminated. Dissemination blocks cover all the territory of Canada. For the 2001 Census, the term 'block' was used (Statistics Canada 2007b).
3. This specification follows the recommended use of the AREAAGGREGATE command: select a cell size that will not cause features to shift too much, yet is not so small that it compromises processing time (Environmental Systems Research Institute 2000, p. 11).
4. Selected ecumene pockets representing important northern communities (such as Inuvik, Iqaluit and Cape Dorset) are buffered to 10 kilometres so that they are visible at small scales.