New approach in modeling spring wheat yielding based on dry periods

The developed model MoDrY is designed for the estimation of the level of spring wheat yields on the basis of dry periods occurring during the particular periods between the phenological phases of the crop plant. A characteristic feature of this model, unlike most existing weather-yield models, is that the principle of its operation is based only on information on the occurrence of precipitation. In the study the authors used research material from 27 years and diurnal sums of atmospheric precipitation. Five measures were adopted to characterise the error of approximation: coefficient of correlation, mean relative error, RRMSE, EF, and CRM. The coefficient of correlation obtained was at the level of 0.91, and the mean relative error at the level of 9.79%. Validation was performed by means of the Cross Validation test (CV), version LOO. For the approximation and validation, the values of CRM and RRMSE indicated a very good fit of the model, also supported by the value of EF obtained for calibration. Additional result of developed MoDrY model is the information that could be helpful to assess the impact of rainless periods occurring during particular interphase periods on the spring wheat yielding.

► We developed the original model, which task is to quantitatively estimate spring wheat yielding.
► The input data are phenological phases of that crop and the number of rainless days.
► Dry periods during the interphase periods II, III, V mainly effect the yielding.
► For the period I, dry periods lasting longer than 4 days are of key importance.
► During period V every single rainless day contributes to the yield increase.