Climate can be understood as the mean atmospheric condition for a certain region. It affects directly most of the human activities, in special agriculture since it defines final yield levels, conditioned mainly by the regional water availability. However, the climatic classifications systems (SCC) are not often used for agricultural studies because usually their scales are considered very wide. Therefore monthly mean values of maximum and minimum air temperature of 27 termometric stations and 427 pluviometric stations of State of São Paulo were utilized to update and improve the mapping of two SCC, Köppen modified and Thornthwaite, besides to evaluate their applicability for agroclimatic zonning for the State of São Paulo. The use of 427 localities resulted in amore accurate mapping of the State by both climatic classifications. Their applicability on agroclimatical studies was evaluated through the separation capacity of the climate types by two systems, in relation to the meteorological elements and components of normal water balance. It was used data dispersion analysis, Tukey´s test and cluster analysis with independent data. The Köppen SCC showed to be efficient only for macroscale and low capacity of climate types separation in relation to meteorological elements (air temperature, rain) and data resulting from water balance (evapotranspiration, deficit and surplus). As a result this SCC should not be used in agrometeorological studies. On the other hand, Thornthwaite SCC efficiently separates the climate types in toposcale or mesoscale in view of the fact that it can efficiently resume generated informations by normal water balance, indicating ability to determine agroclimatical zones. Discussions were also made about sowing dates and final quality of agricultural products related to the two SCC considered question.
agrometeorology; zonning; cluster analysis
