Frost forecast in Santa Catarina State, Brazil

Abstract

Frost is a weather event characterized by the formaton of ice on exposed objects and plants, being observed mainly from May to September in Southern Brazil. Climate study and frost forecast are important for the agricultural sector
due to crop damage and consequent economic losses. This paper proposes a probabilistc model for the forecast of frost in different intensites in the State of Santa Catarina, Southern Brazil. The methodology uses a generalized linear model to determine a frost index, corresponding to the relatonship between the occurrence of frosts in different intensites and the following variables of prognostc Weather Research and Forecastng (WRF) model: minimum daily air temperature (Tmin); relatve humidity and wind velocity, observed at the same tme as Tmin. The results obtained from the data analysis of a 3-year
period showed good performance of the model in predictng frost, especially in the separaton of the different intensites of the phenomenon.