Heat islands in Florianópolis (SC)

a comparative analysis of spatial modeling methods

Abstract

Urban heat islands (UHIs) are a reality in Brazilian cities yet remain underexplored in coastal and insular environments. This article presents a characterization of UHI in Florianópolis and discusses the potential and limitations of spatial modeling techniques (kriging, inverse distance weighting, and multiple linear regression) in contexts with limited sampling. Hourly data from 12 weather stations were used, and the models were applied to a day with stable atmospheric conditions. The results revealed a classic nocturnal UHI pattern, with seasonal influences. Among the spatial modeling methods tested, multiple linear regression provided the most accurate results, considering the geographical particularities of the study area.