Comparing annual wheat yield sensitivity to climate at different sites using impact response surfaces.

Abstract

Impact response surfaces (IRSs) are plots that show the response of a dependent variable (the surface) with respect to two predictor variables. These have been used in recent studies to display wheat yield sensitivity to climate at sites in Europe across an ensemble of crop models. Results focused on period-averaged responses to a wide range of temperature and precipitation perturbations. However, these averaged responses may mask more complex year-to-year sensitivities.In this study we have used the IRS approach to investigate the sensitivity of wheat yields to short-term (inter-annual) climate fluctuations for sites in Finland, Germany and northern Spain. We focus on the baseline period (1981-2100) in order to gain insight into model behaviour in response to present-day seasonal weather variations. We also explore the use of IRS plots to examine regional yield statistics with respect to temperature and precipitation over the same period. Analysis of the IRS patterns allows for a comparison of model behaviour in response to long- and short-term climatic fluctuations as well as a reality check of modelled versus observed yields.Preliminary results indicate some consistency in the association between annual observed yields and temperature and precipitation anomalies compared to long-term responses reported previously. At all sites the relationship between grain yield and climate is positive for precipitation and negative for temperature (except for cool years in Finland). Ongoing analysis is also investigating IRS patterns of inter-annual responses from models participating in an earlier ensemble inter-comparison exercise (Pirttioja et al. 2015) from which co-authors are gratefully acknowledged.