Simulation of yield decline as a result of water stress with a robust soil water balance model

Abstract

The relative yield decline that is expected under specific levels of water stress at different moments in the growing period is estimated by integrating the FAO Ky approach [Doorenbos, J., Kassam, A.H., 1979. Yield response to water. FAO Irrigation and Drainage Paper No. 33. Rome, Italy] in the soil water balance model BUDGET. The water stored in the root zone is determined in the soil water balance model on a daily basis by keeping track of incoming and outgoing water fluxes at its boundary. Given the simulated soil water content in the root zone, the corresponding crop water stress is determined. Subsequently, the yield decline is estimated with the Ky approach. In the Ky approach the relation between water stress in a particular growth stage and the corresponding expected yield is described by a linear function. To account for the effect of water stresses in the various growth stages, the multiplicative, seasonal and minimal approach are integrated in the model. To evaluate the model, the simulated yields for two crops under various levels of water stress in two different environments were compared with observed yields: winter wheat under three different water application levels in the North of Tunisia, and maize in three different farmers’ fields in different years in the South West of Burkina Faso. Simulated crop yields agreed well with observed yields for both locations using the multiplicative approach. The correlation value (R2) between observed and simulated yields ranged from 0.87 to 0.94 with very high modeling efficiencies. The root mean square error values are relatively small and ranged between 7 and 9%. The minimal and seasonal approaches performed significantly less accurately in both of the study areas. Estimation of yields on basis of relative transpiration performed significantly better than estimations on basis of relative evapotranspiration in Burkina Faso. A sensitivity analysis showed that the model is robust and that good estimates can be obtained in both regions even by using indicative values for the required crop and soil parameters. The minimal input requirement, the robustness of the model and its ability to describe the effect on seasonal yield of water stress occurring at particular moments in the growing period, make the model very useful for the design of deficit irrigation strategies. BUDGET is public domain software and hence freely available. An installation disk and manual can be downloaded from the web.