Several theoretical probability distributions, including Normal, Gumbel, Fréchet, Weibull, Lognormal, and Pearson 3 were fitted to annual extremes of daily maximum rainfall, maximum and minimum absolute temperature, and instantaneous wind speed in Piracicaba, São Paulo, Brasil (1917-2006). Maximum likelihood methods were used to estimate the distributions parameters as well as their confidence intervals. KS and χ2 tests were used to compare the fits and select the best theoretical distribution. The Gumbel distribution was that best fitted the extremes of daily maximum rainfall and maximum absolute temperature. The minimum absolute temperatures presented a symmetrical distribution and were best fitted by the normal distribution. The Weibull distribution best fitted the maxima wind speeds. Results showing the expected values, as well as their confidence intervals, for return periods up to 500 years, are presented.
Extremes distributions fittings; meteorological variables; Piracicaba
