Measuring place rather than space: integrating landscape qualities for the urban-rural gradient using people's perceptions and machine learning

This study applies a generalisable method for measuring place across a large urban-rural landscape by combining subjective and objective data for 7,008 locations in the Utrecht region, the Netherlands. A representative online survey (N = 10,042) was used to calculate people's subjective valuations through street view images. The objective measures were obtained from existing national-level data platforms. Regression models, dimensionality reduction and clustering algorithms were used to uncover patterns between perception (subjective) and spatial (objective) components such as land use, accessibility to services or low level visual features. The results show that perception of spaces is a fundamental component of place and people's subjective valuations can be directly combined with other spatial indicators to produce a more nuanced understanding of complex urbanisation patterns. We close by illustrating how place rather than space can inherently characterise human-environment interactions and connect local and landscape scales to navigate the homogenization and degradation processes of periurbanisation.