Referenced by 4 other explainers
Tessellation is the method Place Forecast uses to move ABS data from larger areas down to mesh blocks, then build it back up to small areas. It distributes data from areas like SA1s and SA2s to each mesh block, then adds the mesh blocks up to the small area you see.
Place Forecast always weights this split by dwellings, not by people. For Census markers, the weight is occupied private dwellings. For the core population data, the weight is census dwellings. Place Forecast does not weight by resident population. The ABS does not publish population at mesh block level, so even population is shared out by dwellings.
This has two limits worth knowing. First, some mesh blocks hold people in non-private dwellings, such as prisons, aged-care homes, defence bases, and student housing. These mesh blocks have few or no private dwellings, so they get little or no share of the count. Their residents still sit in the wider-area total, but the chart cannot show them at a finer level. Second, tessellation spreads a small group evenly by dwellings. If a group clusters in just a few streets, the chart spreads it across the wider area. It shows the wider area's pattern, not where the group really lives.
Each mesh block is matched to a small area based on where their boundaries overlap. Place Forecast can split a mesh block across two or more small areas. So small area boundaries do not have to follow mesh block edges. Users can define small areas in any shape they need, and the system cuts mesh blocks to fit. Mesh blocks also link to SA1s and SA2s through their ABS codes rather than boundary overlap. In all cases, no person or dwelling is counted twice, and none is left out.