Simulation of maize yield in current and changed climatic conditions: Addressing modelling uncertainties and the importance of bias correction in climate model simulations

Abstract Appropriate knowledge and understanding of the impact of climatic variability on agricultural production is essential for devising an adaptation strategy. Climate change impact studies have to cope with the cascade of uncertainties that enter at different levels of modelling (e.g., emission scenario, climate model structure, impact assessment models). Our study aims at addressing these uncertainties through an ensemble probabilistic approach, which accounts for uncertainties in climate model simulations as well as parametric uncertainties in a dynamic crop model, when simulating maize ( Zea mays L.) growth and development. Simulations from eight regional climate models were used in combination with 10,000 different parameter sets from a dynamic crop model, reflecting biophysical uncertainties. Since regional climate model simulations can be subject to systematic biases, the use of such simulations to create impact assessment models can lead to unrealistic results. In the second phase of our study, we therefore determined the importance of bias correction of simulated meteorological variables prior to their use as input data in a dynamic crop model. The results revealed that using raw simulations from regional climate models to force a dynamic crop model produced unrealistic maize yield estimations, mainly because of underestimation of the intensity of daily precipitation. Corrected simulations from climate models significantly improved the quality of maize yield simulations, while a lower degree of improvement was observed in cases in which the frequency of wet days was underestimated in comparison to measured values. Using bias corrected climate model simulations in an ensemble probabilistic approach resulted in probability distributions of expected yield changes at three locations in Slovenia. Yield is expected to decrease on average between 10% and 16% in the 2050s and between 27% and 34% in the 2090s, while inter-annual variability is expected to increase compared to the control period between 1961 and 1990. Variance decomposition of the ensemble yield projections was performed in order to determine the RCM inter-model variability and crop model parameter uncertainty. The proportion of variance between RCMs increases during the 21st century, but never exceeds the inter-annual yield variability. Moreover, the parametric uncertainty of the WOFOST model can be regarded as negligible compared to RCM inter-model variability and yield inter-annual variability. A statistical emulator of the dynamic crop model was developed in order to analyze the impact on maize yield of weather variability within the growing season. It has been recognized that maize yield depends mostly on weather conditions during the period from 90 to 110 days after sowing, which coincides with the silking and tasseling period. High temperatures, low relative humidity and low rainfall during this period negatively affect maize growth, leading to a decrease in dry matter production. The analysis also revealed that precipitation during the growing season had a decisive impact on inter-annual yield variability at the selected locations.