A low-cost, low-tech methodology for population estimation in heterogeneous, data-scarce urban settings, applied in the mountainous area of Qwaqwa, South Africa

Accurate population data is essential for effective planning for service and resource allocation, urban development and climate change adaptation – particularly in mountainous regions, where complex topography poses challenges to both data collection and infrastructure development. This paper presents a low-cost, low-tech and replicable hybrid methodology to estimate the population size and spatial distribution and applies it in Qwaqwa, a semi-rural, semi-urban mountainous region in South Africa. Using high-resolution satellite imagery, was manually mapped 91,624 main residential buildings. The study area was stratified into seven urban structure types based on morphological attributes, with 917 randomly sampled sites surveyed, with a response rate of 52.7%. Survey data provided occupancy rates, which were used to calculate yard-level population estimates and extrapolate population density across strata. The study reveals high variability in residents per yard, with no strong relationship between population density and visible physical attributes of main buildings and yards. The assessment resulted in an average simulated population of approximately 360,000 inhabitants (range 320,000–400,000), with a mean of 3.94 inhabitants per main building, including uninhabited buildings. The distribution of the population in Qwaqwa reflects the spatial imprint of apartheid-era planning, topographic constraints and ongoing urbanisation pressures, which together shape where people live and the level of demand placed on local infrastructure and services such as water access, transport and public facilities. The findings highlight the limitations of traditional remote sensing methods in mountainous terrain, where settlement patterns are shaped by both complex topography and socio-economic dynamics.