3.2.7. The Relationship Between Imperviousness and Census Variables

By comparing percent imperviousness as generated for the enumeration area aggregates with census variables, potential relationships can be investigated. A straightforward means of doing so is through the use of correlation measures. Results are presented in Table 10. Correlations between imperviousness and several census variables are calculated on both a per pixel and per polygon basis. The per pixel correlation is related to the per polygon correlation in an areal weighted sense. While both examine the relationship, the per pixel approach will give greater weight to larger enumeration areas, while the per polygon approach is size independent.

Imperviousness is highly correlated with both population and dwelling density. This finding supports an hypothesis that the degree of development is expressed coincidentally through the presence of people and impervious surfaces. It is not surprising that both population and dwelling densities have similar levels of correlation, since the two are themselves highly correlated. The higher coefficients in the per pixel table suggest that the large enumeration areas, where population and imperviousness are generally low, are large contributors to the correlation. The moderately strong negative correlation between imperviousness and percent population with post-secondary degrees suggests that level of education may imply lots with larger areas and more open spaces, since the correlation is significantly smaller in the per polygon case. The remaining correlation coefficients are not significant, and cannot be meaningfully interpreted.