To design hydraulic structures, a sound estimation of rainfall parameters is necessary, in order to ensure the technical and economical feasibility of the project. Equations to predict intense rainfall are widely used to estimate runoff peak discharge. The difficulty to obtain estimates for sites where no rainfall data are available can be overcome by the interpolation of precipitation data, based on neighboring meteorological stations, through statistical procedures. The objective of this work was to compare the two most accurate statistical methods, previously confirmed in other applications, to interpolate rainfall parameters: kriging (geostatistical interpolator) and the inverse-square-distance, for the interpolation of equation parameters of intense rainfall. Data from 140 meteorological stations in Sao Paulo State, Brazil, were used. Each station has its estimated equation of intense rainfall, which is considered ideal, based on historical data series. Spatial dependence of the parameters was tested for 126 of the stations, whereas the other 14 were used to compare the mean errors of the two methods. Both allowed a satisfactory estimation of rainfall parameters. The kriging method can be indicated as the most appropriate since it produced lower mean errors at 11 stations, especially for the return period and duration time commonly applied to estimate rainfall parameters for projects in small watersheds.
intense rainfall; regionalization; geostatistics; hydrology
