The Digital Life of Parisian Parks: Multifunctionality and Urban Context Uncovered by Mobile Application Traffic

TL;DR

The study introduces antenna-azimuth refined mobile traffic attribution to quantify park-specific smartphone activity in Paris, addressing gaps where park use was inferred only from coarse metrics. By processing 492 million hourly records from 4G/5G networks and 41 apps, the authors reveal three functional park types—Cultural, Lunchbreak, and Recreational—each with distinct temporal and app-use signatures. They test the Central-City Multifunctionality and Socio-Spatial Differentiation hypotheses through correlations with neighborhood socioeconomic indicators, finding evidence for both: central parks show higher weekday use and app diversity in wealthier, more unequal areas, while suburban parks reflect local digital cultures. The work demonstrates that passively collected mobile data can inform urban park planning, health promotion, and equity, offering scalable, fine-grained insights into how parks function within complex urban contexts.

Abstract

Urban parks support public health, but landscape architecture typically examines them through form and function. Prior equitable access research focused on park form, while functional studies relied on small-scale surveys, movement data, or broad usage metrics, missing specific activities and visit motivations. This gap limits our grasp of parks' functional diversity. We address this with a novel method refining mobile base station coverage via antenna azimuths to isolate park-specific traffic from surroundings. Using Paris as a case study, we process 492 million hourly per-app mobile records (35% market share) from 45 urban parks. We test the central-city hypothesis (multifunctional parks in dense, high-rent zones due to land constraints) and socio-spatial hypothesis (parks reflecting neighborhood routines and preferences). Results reveal parks' unique mobile traffic signatures, distinct from urban contexts and each other. Clustering by temporal and app patterns identifies three types: lunchbreak, cultural, and recreational parks, linked to health-promoting visitation motives. Central parks show diverse apps and peak usage; suburban recreational parks mirror local demographics, like income-aligned app preferences. This demonstrates mobile traffic's power as a proxy for urban green space activities, with key implications for park design, public health, and well-being strategies.

Figures (4)

• Figure 1: Top: Methodology for selecting parks that can be reliably analyzed. (A) Considering a base station, we leverage the coverage sectors to allow a finer grain Voronoi geometries, which allows a selection of geometries with better overlap with parks. Example of (B1) good coverage (top - Jardin de Luxembourg) vs. (B2) poor coverage (bottom - Square Samuel Rousseau); (C) Plotting park area (x-axis) against the Quality of Coverage $Q_{p}$ (cf. \ref{['subsec:park_selection_methodology']}, y-axis). Larger parks generally have better coverage, leading to a focus on bigger parks. From 1.641 named parks in Paris, we selected 45 (min: 1.2 ha, median: 25.5 ha, max: 995 ha) for our study. Bottom: Traffic analysis across parks. (D) RSCA distribution in parks and the city, indicating app popularity (RSCA $>$ 0) or unpopularity (RSCA $<$ 0). Markers show means, lines indicate $95\%$ confidence intervals. $*$ and $**$ mark significant differences. (E) Ratio of traffic on workdays vs. weekends for individual parks (gray), park average (green), and city average (black). Values $>1$ indicate higher traffic on workdays.
• Figure 2: (A) Park types across Paris. (B) Median weekly traffic per cluster for parks within each cluster. (C) RSCA by application and park cluster. Points show mean values, while lines indicate $95\%$ confidence intervals across parks in each cluster.
• Figure 3: Relation between weekday to weekend traffic ratio (top) and RSCA (bottom) versus socioeconomic indicators, for park categories. Pearson correlation coefficients $\rho$ and $p$-values (in parentheses) are indicated on top.
• Figure 4: RSCA across apps per park category, split between weekday and weekend. Points show mean values, while lines indicate $95\%$ confidence intervals across parks in each cluster.