A1200
Title: A Bayesian ensemble projection of climate change and technological impacts on future crop yields
Authors: Dan Li - Queensland University of Technology (Australia) [presenting]
Vassili Kitsios - Commonwealth Scientific and Industrial Research Organisation (Australia)
David Newth - Commonwealth Scientific and Industrial Research Organisation (Australia)
Terry OKane - Commonwealth Scientific and Industrial Research Organisation (Australia)
Abstract: The purpose is to introduce a Bayesian hierarchical model embedded within a fully Bayesian probabilistic framework. The proposed approach is explicitly designed for rigorous estimation, model selection, and comprehensive forecasting of uncertainties in crop production and, consequently, food security under multiple greenhouse gas concentration pathways. By extending an existing parsimonious multivariate econometric model, the approach explicitly accommodates heterogeneity in the error-term assumptions, replacing the previously restrictive assumption of homogeneity across different countries. Adopting a fully probabilistic method allows us to systematically characterize ensemble uncertainties, capturing a wide spectrum of plausible climate scenarios and technological improvements. Through detailed in-sample and out-of-sample analyses focused on wheat production, we demonstrate the robustness and superiority of our model in quantifying these uncertainties. The enhanced probabilistic insights generated by the model directly inform risk-aware agricultural and climate policies, empowering decision-makers to better prepare for a range of possible futures and implement more resilient and adaptive strategies in the face of climate-related challenges.
