Comparative study of fifty potential evapotranspiration models for urban area in tropical region

Abstract

Evaluation of potential evapotranspiration (PET) models against Penman Monteith FAO-56 has become more popular in developing countries. However, it requires complete weather variables data. To overcome this problem, we evaluated a number of PET models which were classified into fourteen-Temperature based models; twelve-Mass transfer based models; twenty one-Radiation based models and three-Combination based models against 20 years-observed pan evaporation data from Subang Jaya meteorological station. Julian day based mean imputation was used to fill the missing data. Tukey's outlier detection method was employed before estimating the PET. The performance of the models were evaluated with percentage of error (% error) in total values, % error in maximum values, % error in minimum values, % error in average values, root mean squared error (RMSE), and the paired t-test of the prediction accuracy. The results showed that Linacre, Meyer, Conchrane-Orcutt, and Kimberly Penman performed better than the other PET models among their categories. The present study also indicated that Kimberly Penman (combination based model) is decided as the best PET model. However, with considering the small difference of RMSE values and number of required weather variables, Conchrane-Orcutt (radiation based model) is recommended for future research and practical hydrological application.