Modelling climate change adaptation in European agriculture: Challenges and priorities.

Abstract

Uncertainty in climate change impact projections originates mainly from the inadequacies in structure and parameters of the impact model, climate change scenarios and other input data. Previous studies tried to account for the uncertainty from one or two of the major sources. Here, we developed a super-ensemble-based probabilistic projection to account for the uncertainties from three major sources comprehensively. We demonstrated the approach by assessing projected climate change impact on barley growth and yield in the Boreal and Mediterranean climatic zones in Europe using eight crop models and multiple sets of crop model parameters under three representative climate change scenarios for the 2050s. We further quantified and compared the contribution of crop model structure, crop model parameter and climate change scenario to the mean squared error using the multivariate analysis of variance. The projected changes in barley yield due to climate change by the 2050s ranged from -45.8% to +26.3% at Jokioinen, Finland and from -54.8% to +78.6% at Lleida, Spain, relative to 1981-2010 level. Based on the super-ensemble probabilistic projection, the median of simulated yield change was -3.8% and +7.5%, and the probability of yield decrease was 0.57 and 0.43 in the 2050s, at Jokioinen and Lleida, respectively. The contribution of crop model structure, crop model parameters, and climate change scenarios to the mean squared error was, respectively, 37%, 53%, and 2% at Jokioinen, and 46%, 40%, and 2% at Lleida, for our setting with just three different climate scenarios. The super-ensemble-based probabilistic approach can provide more useful information and better understanding of the uncertainties in climate change impact projections. and stakeholder interaction (e.g. communication with, relevance for). For operational and tactical strategies (changes in practice in response to daily, monthly, or seasonal variation in conditions) most challenges were technical, relating to limitations in the processes and mechanisms represented in models. For longer term strategic climate change adaptation, uncertainty about future socio-economic context (e.g. prices and regulation) and the impact of new adaptation options (e.g. appearance of new technologies) were highlighted. Progressively novel and far-reaching strategies increasingly challenge the scope of existing models. Whilst models vary in capacity, most modellers reported a potential to better characterise adaptation. However, costs (e.g. trade-offs with processing speed) and the fact that adaptation lies beyond the initial remit of many models mean that strategic prioritisation of adaptation as a focus for modelling is key to facilitating model development to support effective stakeholder choices.