Performance Evaluating of Combination Techniques for Enhancing Water Requirement Estimation of Saffron (Crocus Sativus L.) in Iran

Abstract

An accurate estimation of the water requirement of saffron, as the most strategic product in the eastern regions of Iran, is inevitable. Considering the field limitations in measuring the water requirement, applying empirical models has always been of interest. However, since each estimation model has unique strengths and weaknesses, relying only on an experimental model cannot obtain a reliable estimate for water requirements. This study intends to evaluate different combined methods' ability to merge the saffron water requirements simulations and obtain an improved output. Six empirical models and four other combination techniques were applied to get some skilful simulations about saffron water requirements in arid regions. Results indicate that the evapotranspiration prediction under the Abtew method (ABM) has more proficiency, such that its RMSE was 0.13 mm. Also, the different comparative tests show that the outputs of combined techniques such as Multi Model Super Ensemble ‘MMSE’ and Modified MMSE ‘M3SE’ outperform others.

Keywords

• empirical evapotranspiration models
• ensemble modeling
• multi- collinearity
• post processing