Machine learning-based evaluation of dynamic thermal-tempering performance and thermal diversity for 107 Cambridge courtyards

Zhikai Peng*, Ramit Debnath, Ronita Bardhan, Koen Steemers

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

2 Citations (Scopus)
79 Downloads (Pure)

Abstract

The dynamic thermal conditions profoundly impact on the quality of physical, cultural, and social experiences in courtyard spaces. This research aims to identify the microclimatic dissimilarities between courtyards in terms of tempering seasonal–diurnal thermal extremes and enriching ground-level thermal textures. The methodology included field measurements in summer-2021 and winter-2022 in Cambridge, UK; microclimatic simulations of 107 courtyards in ENVI-met and model validations; and machine learning-driven clustering using Super Organising Maps (SuperSOM). The results indicate that the diurnal thermal range of the spatial-UTCI mean in summer (DTR(M)<24C) is double that in winter (DTR(M)<12C); meanwhile the maximum spatial-UTCI deviation is three times as significant (δ>3Cat 7:00 BST versus δ>1Cat 12:00 GMT). SuperSOM analysis was performed using K-means and hierarchical agglomerative clustering to partition all courtyards into seven subclusters on its graph-lattice structure. Clusters Km_I, Hac_I, and Hac_IV feature a positive synergy between the thermal-tempering and thermal-enriching potentials. In contrast, the other four clusters exhibit conflicting scenarios during the day and night across the two seasons analysed. These data-driven outcomes enabled us to optimise spatial and landscape strategies for designing and retrofitting courtyard microclimates, contributing to the current discussions on climate-responsive and sensation-inclusive design in historical urban contexts.

Original languageEnglish
Article number104275
JournalSustainable Cities and Society
Volume88
DOIs
Publication statusPublished - 2023

Keywords

  • Courtyard
  • Historical urban contexts
  • Machine learning
  • Microclimate
  • Thermal diversity
  • Thermal tempering

Fingerprint

Dive into the research topics of 'Machine learning-based evaluation of dynamic thermal-tempering performance and thermal diversity for 107 Cambridge courtyards'. Together they form a unique fingerprint.

Cite this