Document Type : Research Paper

Authors

• Mahsa Sameti ¹
• NOZAR GHAHREMAN ²
• Khalil Ghorbani ³

¹ M.Sc student, Deptment. of Irrigation and Reclamation Engineering, University of Tehran;

² Associat. Professor., Deptment. of Irrigation and Reclamation Engineering, University of Tehran, Karaj;

³ Faculty Member of organ University of Agricultural Sciences and Natural Resources

Abstract

The capability of M5 model tree in estimating reference evapotranspiration in two stations, namely, Shiraz and Kermanshah, was studied. The daily weather data including mean air temperature, sunshine hours, precipitation, dew point temperature, mean relative humidity, wind speed and actual vapor pressure were collected and used as input variables for daily estimation of potential evapotranspiration by Penman-Montieth and Hargreaves-Samani equations. As the main goal of this study, the performance of M5 model tree in predicting reference evapotranspiration was evaluated. The results showed that the skill of M5 model in predicting ET by both methods was high but its performance in predicting Penman-Montieth values was relatively more, with R² of 0.975 and 0.973 and RMSE values of 0.346 and 0.361 for Shiraz and Kermanshah, respectively, while the corresponding values for Hargreaves-Samani equation were Shiraz:R²=0.837, RMSE=0.844, and Kermanshah:R²=0.979, RMSE= 0.774. Sensitivity analysis revealed that the most significant variables in Penman-Montieth equation in Shiraz station were, respectively, air temperature, dew point, sunshine hours, and wind, while in Kermanshah station, important factors were air temperature, sunshine hours, wind, relative humidity, and dew point, respectively. Further studies are recommended in other climates for more scrutiny.

Keywords

• Sensitivity analysis
• Penman- Monteith equation
• Data mining
• Hargreaves- Samani equation