عنوان مقاله [English]
Application of crop models is essential for numerous purposes including prediction of crop yield and water requirement, evaluation of the impact of climate change, drought and irrigation and agronomic management on crop growth and development. One challenge with the application of these models is the large number of input parameters. Measurement of input parameters can be time-consuming, costly and sometimes practically impossible and they are usually estimated using calibration and inverse modelling. Sensitivity Analysis is a procedure during which the impact of input parameters on target output variables is investigated. During model calibration, sensitive parameters must be measured or estimated with higher accuracy. Analysis of the sensitivity of closed-source models is not as straightforward as it is with open- source models. In this research, the impact of 47 crop parameters on five output variables of AquaCrop, a closed-source crop model, namely soil evaporation, crop transpiration, evapotranspiration, biomass at harvest and grain yield, were studied for wheat and maize in Qazvin Plain and Moghan Pars-Abad in Iran. The sensitivity of the selected parameters was evaluated with the relative Nash–Sutcliffe Efficiency Index. Increase in canopy cover, degree-days from sowing to maturity, degree-days from sowing to start of senescence and maximum canopy cover in fraction of soil cover were identified as sensitive parameters for both crops. Therefore, accurate determination of crop growth stages, in calendar days or degree-days, which are easier to measure than most of the other parameters, is of greater importance. In Qazvin and for all output variables, around half of the selected parameters were ineffective and considered unimportant during calibration. Despite that the methods of local SA are computationally and conceptually simpler than the methods of global SA, the results showed that this method could lead to similar outcomes to previous studies in which global methods were used.