Abstract
Climate classification plays an important role for the identification of homogeneous groups of climates, from which representative locations can be extracted and used for building energy simulation analyses. Nevertheless, according to the current state-of-the-art, the main reference systems consider just a fraction of those weather quantities which are relevant in the building energy balance, i.e., ambient temperature and humidity and solar radiation. To overcome this issue, in previous researches a new methodology was defined, based on monthly series of weather quantities, statistical analyses and data-mining techniques for climate clustering. In this work, with the aim of further developing such approach, a shorter time-discretization of weather quantities, i.e., a weekly discretization, was tested, alongside additional variables describing the daily range of ambient temperature and humidity. In order to investigate the potential of those modifications, a dataset with more than 300 European reference climates was analyzed and subdivided into climate classes according to the proposed clustering procedure.