Built Environment, Active Travel, and Location-Based Outdoor Activity: Spatial and Temporal Perspectives

Outdoor physical activity plays an essential role in fostering public health, social well-being, and environmental sustainability. Yet despite its widely acknowledged benefits, participation in outdoor physical activity remains insufficient, both globally and in the Netherlands, where over half of the population does not meet recommended guidelines. Sedentary lifestyles, changing work patterns, and increasing dependence on technology continue to intensify this challenge. Against this background, the built environment has been recognized as a crucial determinant of outdoor physical activity, shaping the opportunities and constraints for individuals’ daily routines. While previous studies have offered valuable insights into how built environment affect outdoor physical activity, they often focused on a single spatial scale, assumed linear relationships, or failed to account for the temporal dynamics of life-course transitions. Moreover, distinctions between different forms of outdoor physical activity, such as conventional cycling versus e-biking or active travel versus location-based outdoor activities, have frequently been overlooked. In addition, lifestyle changes such as telecommuting have introduced new behavioral patterns that remain underexplored. This dissertation addresses these gaps by developing an integrative framework that combines spatial and temporal perspectives in the study of outdoor physical activity. The research focuses on three behavioral outcomes: cycling volume, duration of location-based outdoor activities, and frequency of active travel including walking, conventional cycling, and e-bike use, and asks how the built environment shapes these behaviors across scales and over time. To do so, it formulates four objectives: to evaluate the associations between built environment characteristics at different spatial levels and cycling volume; to examine how location-level and neighborhood-level features relate to the duration of outdoor activities; to investigate how life events and built environment jointly influence active travel frequencies across time, considering lead, concurrent, and lagged effects; and to explore how telecommuting frequency and related behavioral changes are connected to active travel. The empirical studies draw on diverse data sources and apply advanced computational methods. In the first study, we combined hourly cycling counts from 279 monitoring locations with Google Street View images, OpenStreetMap data, POIs, digital elevation models, and Statistics Netherlands data, to study the possible nonlinear associations between multi-level built environment characteristics and cycling volumes using Gradient Boosting Decision Trees (GBDT). In the second study, GPS trajectory data from the B-Riders program, recording nearly 3,000 participants’ activities in Noord-Brabant, were integrated with multi-level built environment characteristics to investigate the duration of location-based outdoor activities, again using GBDT to capture complex, non-linear relationships. In the third study, we used longitudinal data from five waves of the Netherlands Mobility Panel (MPN) and a Dynamic Bayesian Network (DBN) to uncover temporal dependencies between life events, the built environment, and active travel frequencies. In the fourth study, a Bayesian Network (BN) was used with MPN data to analyze the role of telecommuting in shaping active travel patterns, distinguishing between walking, conventional cycling, and e-bike use. The results show that cycling volume is strongly associated with both neighborhood and street-level characteristics, with population density, residential density, bus stop density, and cycleway density emerging as particularly influential. Importantly, the relationships are not linear: for example, cycling increases with density only up to certain thresholds, beyond which the effect weakens or reverses. Weather also plays a crucial role, with moderate temperatures and low humidity encouraging cycling. The study of location-based outdoor activities similarly reveals that greenery, water features, and land-use diversity support longer durations of outdoor activity, but only after certain thresholds are reached. Steep slopes, in contrast, act as a deterrent. The temporal analyses provide new insights into the dynamics of active travel across life events. The DBN approach shows that household relocation, employment changes, and other transitions have not only concurrent but also lead and lagged effects on walking, conventional cycling, and e-biking. In particular, e-bike adoption follows distinct temporal patterns, often substituting for walking rather than complementing it. The analysis of telecommuting highlights its double-edged nature: while remote work reduces commuting trips, it can generate additional opportunities for walking and cycling for non-work purposes. At the same time, e-bike use appears negatively associated with frequent walking, suggesting substitution effects between modes. Together, these findings demonstrate that lifestyle shifts such as telecommuting reshape the landscape of active travel in nuanced and sometimes unexpected ways. The dissertation makes several academic contributions. By integrating neighborhood-, street-, and location-level indicators, it demonstrates that outdoor physical activity is simultaneously shaped by environmental features at multiple scales. By applying machine learning models, it captures non-linearities, thresholds, and interactions that linear approaches overlook, yielding more robust and policy-relevant insights. By adopting a life-course perspective, it provides one of the first systematic analyses of lead, concurrent, and lagged effects of life events on active travel, highlighting dynamic behavioral adaptations. Finally, it contributes novel evidence on the nexus between telecommuting and mobility, with implications for both transport and public health. Policy implications arise from these findings. Promoting cycling requires attention to optimal ranges of density and infrastructure provision, rather than assuming that more is always better. Designing environments that extend the duration of outdoor activities involves investment in green spaces, water features, and mixed land use, while recognizing the thresholds at which these features become effective. Policies aimed at fostering active travel should also consider life transitions as critical intervention points, offering targeted measures during key moments such as household relocation or employment change. With the rise of telecommuting, urban planners must anticipate shifts in travel demand and create environments that encourage walking and cycling for both work and non-work purposes. The research also acknowledges its limitations. While probabilistic models reveal complex associations, causal inference remains constrained, particularly in cross-sectional designs. The findings are based on Dutch datasets and may not be directly generalizable to other contexts with different cultural, infrastructural, or climatic conditions. Some social and psychological dimensions of activity, as well as distinctions between different types of e-bikes, remain underexplored. Future studies should address these limitations through causal designs, cross-country comparisons, and broader integration of social and behavioral factors. In conclusion, this dissertation demonstrates that outdoor physical activity is shaped by a complex interplay of built environment features, temporal life-course dynamics, and lifestyle changes such as telecommuting. By bridging spatial and temporal perspectives with advanced analytical methods, it provides a more nuanced and comprehensive understanding of how walking, cycling, and location-based outdoor activities are embedded in the urban fabric. The findings contribute both theoretically and practically to the pursuit of healthier, more sustainable, and more resilient mobility systems.