This study aimed to evaluate the performance of the models CLIGEN, LARS-WG and PGECLIMA_R in the simulation of daily series of maximum air temperature for localities in the state of Parana, Brazil. It was used time series of maximum air temperature in the localities of Campo Mourão, Castro, Curitiba, Ivaí, Londrina, Maringa and Paranaguá. The validation process consisted of statistical analysis using significance tests (on the monthly average (t test), on the variance of monthly average (F test), on the shape of frequency distributions (K-S test)), of annual trend graphics, of tables with the "r" index of Pearson, "d" index of Willmott and "c" index of Camargo-Sentelhas. In the simulation of maximum air temperature the PGECLIMA_R and LARS-WG generators were not different in performance, obtaining fewer rejections and good results for the "c", "d" and "r" indexes. However CLIGEN obtained the worst results, overestimating the maximum temperatures on wet days and underestimating them on dry days in the localities evaluated in the state of Parana.
weather generators; climatological statistics; Parana climate; synthetic climate series
