Modelling transpiration, soil evaporation and yield prediction of soybean in North China Plain

The main objectives of this study were to assess and partition soybean evapotranspiration and modelling to predict yields. The SIMDualKc water balance model, that adopts the dual crop coefficient approach, was used to evaluate the transpiration and soil evaporation components. Transpiration estimates were then used with the Stewart's water-yield model to predict soybean yields. SIMDualKc was calibrated and validated using soil water observations relative to four crop seasons and six treatments. In addition, the adopted soil evaporation approach using the Ritchie's model was validated against microlysimeter observations, also for the four years of study. The calibrated Kcb was 1.05 for the mid-season and 0.35 at harvesting. Model results show a good agreement between available soil water data observed and predicted by the model, with root mean square errors of estimates (RMSE) smaller than 5% of the total available soil water. Testing the soil evaporation approach also produced good fitting results, with RMSE averaging 0.50mmd−1, hence confirming the appropriateness of the Ritchie's model to estimate soil evaporation of a cropped soil. The yield prediction through combining SIMDualKc and the Stewart's model was successful for all treatments, leading to a small RMSE of 381kgha−1 representing less than 11.5% of the maximum observed yield. These results indicate that yield may be predicted with that simple empirical approach provided that transpiration is accurately estimated and the water yield factor Ky is adequately calibrated. Consumptive water productivity WPET were high, ranging 0.95–1.46kgm−3, showing that both the crop variety and the agronomic practices may be extended in North China Plain.