Improving estimates of occupancy rate and population density in different dwelling types

Population density is heterogeneous, and using large spatial areas as a basis for estimates from highly urbanised areas leads to unrepresentative values. This work shows that population density estimated at the census district level (average 225 dwellings) in Canberra, Australia, poorly reflects dwelling types. Data at the individual block level (net or gross block area) greatly improve the estimates. Eight typical dwelling types in Canberra are used to show that there is a relationship between building form and estimated population density only when population density is calculated using the ‘net block’ area. To estimate population density at a finer scale than census district, the number of occupants in individual dwellings must be estimated. Assuming a city-wide constant occupancy rate in all dwelling types results in a twofold overestimation of population density in high-density dwellings. Fitting a polynomial function to the occupancy-rate and block-area data for different dwelling types of the city also provides a closer estimate than a categorical (step-wise) estimate; the occupancy rate estimate is then easily calculated from a single variable, the mean gross block size in the census district where the dwelling is located. In high-density dwellings in Canberra (e.g. more than 10 storeys), the occupancy rate was approximately 1.3 people per dwelling and in low-density dwellings (e.g. > 1000 m 2 per dwelling) the occupancy rate approached 2.8 people. This work is of value to researchers and planners who use measures of population density for assessing, for example, the per capita resource sustainability of different buildings.