Agricultural projects which deal with hydraulic projects and do not possess observed data on outflow need to explore at the most, information about the Intensity-Duration-Frequency (IDF) curves. Thus, it is necessary to create ways to develop methodologies that estimate IDF curves for locations that have little or no pluviometric data. The aim of this work was to compare disaggregation methodologies for daily precipitation, to verify the increase in quality information considering the IDF curves, as compared to those originated from observed data (historical records). Four methods were tested: (a) Method of Relations (CETESB, 1979); (b) one proposed by BELTRAME et al. (1991); (c) one proposed by ROBAINA & PEITER (1992), and (c) the Bartlett-Lewis Model of Modified Rectangular Pulse (DAMÉ, 2001). A data set for daily precipitation in Pelotas-RS for the period of 1982-1998 was used, and to estimate IDF curves through historical records six time periods were established: 15; 30; 60; 360; 720 and 1,440 minutes, together with return periods of 2; 5 and 10 years. Values for maximum intensities were compared through the Student "t" test for the linear and angular coefficients, and through the relative mean square error. The method that best simulated maximum precipitation intensity for the return periods of 2 and 10 years was the Method of Relations (CETESB, 1979).
maximum precipitation; relationship method; daily rain disaggregation
