The spatial configuration of local climate zones reveals effects on wayfinding in human walking

2.3 Wayfinding measurements 2.3.1 Intelligibility as the proxy for wayfinding. In space syntax context, wayfinding is quantified by an indicator of intelligibility. Intelligibility, measured as the numeric value, refers to the level of coherence between small- and large-scale urban spatial configuration [25], that is, the extent to which large-scale spatial characteristics of a city can be predicted by small- scale spatial features. Spaces with high intelligibility values have more similar and coherent spatial configurations between local and city levels. People are thus easier orient themselves in a complex built environment by predicting large-scale spatial configurations based on small- scale spatial information within their viewshed [8]. [26] reported the first study to relate way- finding directly with intelligibility metrics following the space syntax theory. After that, there have been numerous applications of intelligibility based on the space syntax theory in indicat- ing wayfinding in human movement [27, 28]. 2.3.2 Quantifying intelligibility indicator using spatial design network analysis. Intelli- gibility quantification was conducted using spatial design network analysis (sDNA) (https:// sdna.cardiff.ac.uk/sdna/). Developed from the space syntax approach, sDNA is an ArcGIS/ QGIS/AutoCAD plug-in or Python/command-line tool for analyzing the urban network spatial patterns by simplifying the complex urban spaces into street network representations [29, 30]. To perform the sDNA, street vector layer data was extracted from OpenStreetMap (OSM)

in the QGIS software. OSM is an open-source project to collect and provide free geographical information around the world [31], which has been fairly accurate currently. Around 80% of the highway objects overlap those in the real world [32]. Studies also showed that OSM has high quality when compared to authoritative datasets [33, 34]. The roads derived from OSM are categorized into seven classes based on their social function and importance, including ‘Motorway’, ‘Trunk’, ‘Primary’, ‘Secondary’, ‘Tertiary’, ‘Unclassified’, and ‘Residential’ classes. The least important ‘Residential’ class, which is used to connect housing within the residential areas [35], was excluded from our study, because people in living quarters navigate mainly based on subjective memory rather than spatial configurations. Then, using the suggested six classes of street data, multiple results of street network patterns were provided after conducting sDNA, such as Network Quantity Penalized by Distance in Radius Angular (NQPDA) and Line Connectivity (LConn). Here, intelligibility values are the coefficient of determination between LConn and NQPDA, ranging from 0 to 1 [36]. NQPDA value can be a proxy for urban spatial accessibility that refers to the ease with which people can reach their destinations Fig 2. The location of the Guangdong-Hong Kong-Macau Greater Bay Area (GBA). https://doi.org/10.1371/journal.pone.0289780.g002 PLOS ONE The spatial configuration for wayfinding PLOS ONE | https://doi.org/10.1371/journal.pone.0289780 September 8, 2023 5 / 18