Applicability of thermoregulation-sensation models and machine learning modelling to simulate dynamic physio-psychological thermal responses during walking in urban continuum

Abstract

Walkability is an important attribute of a liveable city, and in this era with frequent heat waves the thermal comfort of walking pedestrians can be essential for the microclimate design of walking routes. Upon field tests conducted during summer in urban continuum in Hong Kong, this study examined the applicability of thermoregulation models, including the Gagge 2-node model and multi-node-segment JOS3 model, both of which are updated with a newly obtained convective heat transfer coefficient, for the accurate evaluation of the dynamic physio-psychological responses of walking pedestrians in the urban continuum. Fiala dynamic thermal sensation (DTS) model was assessed for its effectiveness in simulating transient thermal sensations during walking. Moreover, the study utilised the random forest (RF), a machine learning algorithm, to model transient thermal sensations and average thermal acceptance during walking and resting in the urban continuum. The results indicate that the 2-node model, the JOS3 model, and the human body differ in key determinants of mean skin temperature, and the Fiala DTS model underestimates the impacts of skin temperature change rate and thermal pleasure on transient thermal sensations. Body mass index (BMI) is an important factor affecting the dynamic physio-psychological responses, which is not well considered in any of the three models. The developed RF models exhibit high accuracy in simulating dynamic physio-psychological thermal responses and overall thermal acceptance over a period of time.