Abstract
Population data in Canada is published against Census Dissemination Blocks, irregular polygons whose shape reflects administrative and privacy constraints rather than the spatial distribution of residents. This paper presents a method for disaggregating Dissemination Block population onto a uniform Uber H3 hexagonal grid, resolution 8, mean cell area 0.74 square kilometres, using dasymetric allocation. The goal is to place population where people plausibly live within each block rather than distributing it evenly across block area or collapsing it to a single centroid. Two independent allocation methods are constructed from open Government of Canada data. The first uses points from Innovation, Science and Economic Development Canada's Pseudo-Household dataset, which distributes population along road segments. The second uses Statistics Canada's National Address Register, which distributes population across real address points. Both methods conserve population exactly against published Dissemination Block totals when aggregated to the provincial level. Comparing the two hexagon layers shows close agreement across the large majority of the country and a smaller set of hexagons where the two methods diverge sharply. We quantify this disagreement with a threshold-free displaced population index, defined as the share of national population that would need to move between hexagons to reconcile the two layers exactly, which stands at 8.5 percent nationally. Diagnostic decomposition of individual divergent hexagons down to the contributing Dissemination Blocks shows that this disagreement has more than one cause: a minority of high-density buildings with no National Address Register coverage at all, and a separate, smaller effect in which the two methods place the same residents into adjacent hexagons differently near block boundaries. We propose a hybrid allocation rule that uses the National Address Register wherever a Dissemination Block has address coverage above a people-per-point threshold and falls back to the road-based method otherwise, which corrects the coverage-gap mechanism while preserving exact population conservation. We present this comparison of two independently constructed dasymetric layers, used to locate and explain disagreement rather than to validate either layer on its own, as a useful diagnostic strategy for hexagonal population products built from administrative open data.