The Food and Agriculture Organization (FAO) of the United Nations has recommended the use of the Penman - Monteith equation to estimate reference evapotranspiration (ETo) in a daily time scale assuming that the grass resistance (rc) to water vapor transfer is constant and equal to 69 s/m, and that the ratio rc/ra = 0.33U, being U the daily mean wind speed (m/s) and ra the aerodynamic resistance. The FAO panel of experts asked for testing of such assumption under different climates and this is the objective of this research project. The experimental area is located at Piracicaba, SP, Brazil (22 42' S; 47 30' W; 520 m), where 70% of the annual rainfall concentrates in the summer - autumn months. Data from a weighing lysimeter (0.92 m^2 of exposed area) and an automatic weather station, both located in a grassy area of 3150 m^2, were used to test the goodness of the model estimates.
Discarding rainny days from the analysis, it was found that, on average: a) during the wet months, when the evapotranspiration represented about 70% of the net radiation, the model overestimated by 12%; b) in the drier months, when over 90% of the net radiation was converted into latent heat, the situation was reversed and there was an underprediction also of 12%. This indicates that the assumption of a constant rc is not valid for such tropical climate, and it had to be increased during the wet period, but reduced in the dry months. It was also found that assuming rc/ra = 1.67 U - 1.67, as proposed elesewhere, improved somewhat the predictions giving a 1:1 fit during the wet period. Again, for the drier months, such approach worsen the estimates for it increased rc and the results show that it had to be reduced to 40 s/m to give an appropriate fit. During the dry period the original Penman equation (rc = 0) performed very well giving a 1:1 fit to measurements; however, during the wet period it resulted in a 26% overprediction.