Florianópolis (SC) temperature analysis using a GAMLSS approach

Abstract

Understanding the variability of climate elements in the temperature is important for economic activities and people's daily lives. With this in mind, the main aim of this paper is to analyse the average temperature of Florianópolis, SC over a one-year period (1 July 2021 to 30 June 2022). The following explanatory variables were considered for this task: date (time), dew point temperature, total precipitation, atmospheric pressure, humidity, and wind speed. The generalised additive models for location, scale and shape (GAMLSS) were used due to their flexibility to explain the behaviour of the response variable. The Box-Cox power exponential (BCPE) distribution was chosen to explain the response since it can deal with positive variables with varying degrees of kurtosis. A stepwise-based method was performed to select covariates in each of the distribution's parameters. The residuals obtained from the final model were found to be adequate for explaining the data set.