Reliable estimation of reference crop evapotranspiration (ET0) is of great importance for irrigation planning and operation as well as for climatologic and hydrologic studies. This study evaluated the performances of pan evaporation-based ET0 estimation methods. First,we detected the correlation between observed pan evaporation (Ep) and ET0 estimated by Penman-Monteith FAO-56 (PMF-56). Second, we estimated ET0 from Ep using six pan evaporation models (Cuenca, Allen and Pruitt, Snyder, modiﬁed Snyder, Pereira and Orang methods) and compared them with ET0 by the PMF-56 method. The accuracy of the models was assessed based on three performance statistics such as R2, mean absolute error, and root mean square error.We used daily meteorological data recorded at the Mymensingh weather station for the period of 20072016 to estimate class A pan coefficients (Kp) using the empirical equations proposed by the selected models. Daily ET0 was then estimated by multiplying the Kp values with the corresponding daily Ep values. Daily EP and ET0 values showed moderate correlation whereas monthly values showed high correlation only for February, August, and September. The moderate correlation between daily values is mainly due to the dissimilar response of Ep and ET0 to their influencing meteorological factors. In estimating daily and monthly ET0, overall all methods showed poor performances with underestimated PMF-56 ET0. However, in the case of August ET0 estimation, we noticed better performances from pan evaporation models in terms of lower errors and high R2 (>0.70). Particularly, the Snyder model ranked first among the selected pan evaporation models as it closely predicted PMF-56ET0. So, after necessary calibration, this method can be considered for the estimation of ET0 under the climatic condition of Bangladesh. To conclude, the findings of this study will be a useful reference for adopting a comparatively easier ET0 estimation method in the country.
Key words: Evapotranspiration, Evaporation, Penman-Monteith method, Snyder method