A stochastic model for the analysis of maximum daily temperature

In this paper, a stochastic model for the analysisof the daily maximum temperature is proposed.First, a deseasonalization procedure based on the truncatedFourier expansion is adopted. Then, the Johnsontransformation functions were applied for the data normalization.Finally, the fractionally autoregressive integratedmoving average model was used to reproduceboth short- and long-memory behavior of the temperatureseries. The model was applied to the data of theCosenza gauge (Calabria region) and verified on otherfour gauges of southern Italy. Through a Monte Carlosimulation procedure based on the proposed model,105 years of daily maximum temperature have beengenerated. Among the possible applications of the model,the occurrence probabilities of the annual maximumvalues have been evaluated. Moreover, the procedurewas applied for the estimation of the return periods oflong sequences of days with maximum temperatureabove prefixed thresholds.