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

3.2 Quantitative spatial configurations in the region The LCZ classification in the GBA has been shown in Fig 3 to illustrate the spatially heteroge- neous built and land-cover types. The built-up urban forms were concentrated in the central region where the intelligibility levels were also generally lower (Fig 5), which implies the asso- ciations between wayfinding and LCZ spatial configurations. Using this LCZ map as an input of the FRAGSTATS tool, the class- and patch-level landscape metrics of each LCZ type in each GBA city can be computed. 3.3 Correlations of intelligibility-based wayfinding with LCZs configurations The correlations between landscape metrics of diverse LCZ types and intelligibility values were investigated to examine whether LCZ spatial configurations have statistically significant influ- ences on wayfinding. The results showed that, at the class level, the value of SPLIT metric in LCZ 2 (compact mid-rise) had a positive impact on intelligibility-based wayfinding with a Table 1. The selected landscape metrics, their descriptions, and potentially qualitative impacts on wayfinding. Patch-level landscape metrics Metrics Abbreviation Description Potentially qualitative impacts on wayfinding Euclidean nearest- neighbor distance ENN The shortest straight-line distance between a patch and its closest neighbor. Decreased ENN value may improve wayfinding. Similarity index SIMI The similarity between adjacent patches in terms of their land property. E.g., grasslands have a higher similarity to trees than the impervious surface. Increased SIMI value may

improve wayfinding. Class-level landscape metrics Metrics Abbreviation Description Potentially qualitative impacts on wayfinding Interspersion & Juxtaposition Index IJI The extent to which a class intersperses and intermixes with other classes. Increased IJI value may improve wayfinding. Percentage of Like Adjacencies PLADJ The frequency with which patches of two different classes are found side-by-side in a region. Increased PLADJ value may improve wayfinding. Aggregation Index AI The extent to which two patches share an edge. Increased AI value may improve wayfinding. Clumpiness Index CLUMPY The extent to which classes are clumped. E.g., classes are maximally disaggregated when CLUMPY equals -1, distributed randomly when CLUMPY equals 0, and maximally aggregated when CLUMPY equals 1. Increased CLUMPY value may improve wayfinding. Landscape Shape Index LSI The degree to which a class is irregular in comparison to a single square patch. Increased LSI value may improve wayfinding. Normalized Landscape Shape Index nLSI The normalized version of the landscape shape index (LSI). Increased nLSI value may improve wayfinding. Patch Cohesion Index COHESION The extent to which the patches of the same class are spatially connected. Increased COHESION value may improve wayfinding. Number of Patches NP The description of the number of patches in a class. Increased NP value may improve wayfinding. Patch Density PD The number of patches of a class per unit area. Increased PD value may improve wayfinding. Landscape Division Index

DIVISION The possibility of two randomly chosen patches from different classes. Increased DIVISION value may improve wayfinding. Splitting Index SPLIT The total landscape area (m2) is divided by the sum of patch area (m2). Increased SPLIT value may improve wayfinding. Effective Mesh Size MESH Perfectly inversely correlated with DIVISION. Decreased MESH value may improve wayfinding. https://doi.org/10.1371/journal.pone.0289780.t001 PLOS ONE The spatial configuration for wayfinding PLOS ONE | https://doi.org/10.1371/journal.pone.0289780 September 8, 2023 8 / 18