A new microclimate zoning method based on multivariate statistics: The case of Reunion Island

Accurate knowledge of climatic conditions is essential for planning and developing a territory, particularly in building construction sector. Determining microclimate zoning could benefit policymakers and building designers in enhancing bioclimatic design and thermal comfort under subtropical conditions. This article presents a new microclimate zoning combining statistical analysis (double nested clustering and K-means partitioning) and GIS-based spatial interpolation . The method was tested on meteorological data from 242 weather data points of Reunion island, and the altitude. The first implementation distinguished three climate zones, most notably separating the east (humid region) from the west (dry region), and a central area (temperate region). The second implementation allows to identify subclimate typologies within the three main climate zones. The representation of the local climate dynamics is portrayed through spatial interpolation. The twelve sub-zones provide a more accurate climatic map of Reunion Island, in tune with observed climates over the territory. Moreover, considering altimetry enabled to include the effect of relief in microclimate zoning. The major result of this study is that our microzoning method is not limited to urban areas, unlike most of the case studies in the literature. By considering meteorological data at a fine spatial resolution, our approach can be applied to a wide range of applications (cities, rural areas, energy, agriculture…). [Display omitted] •A double hierarchical bottom-up classification for microclimate zoning is proposed.•It allows multivariate analyses to be performed on diverse climate data.•Updated boundaries of microclimate map for Reunion Island is presented.