Statistics Seminars: A Bayesian calibration of a soil capital production function integrating two types of data
22 April 2016 14:00 in CM221
The amount of soil organic carbon influence the yield effect of nitrogen applied on a field and can be seen as a natural capital in the soil. Quadratic productions functions are currently being used to evaluate the impact of managing the soil capital and to find optimal nitrogen loads in agent based modelling of farmer’s decisions. I will talk about our procedure to calibrate a simplified version of the soil capital production function with two types of data, long-term field experiments and yearly summary statistics from the production region for which the function will be applied. The calibration is performed as a Bayesian evidence synthesis implemented by Markov Chain Monte Carlo simulations in jags in R. The influence of summary statistics data relative to the long-term field experiments is controlled by, a somewhat arbitrary weight, assigned to the precision of summary statistics. I perform robust analysis to study how differences between summary statistics and long-term studies influences the robustness in optimal nitrogen loads derived by corresponding calibrated production functions. The calibrated production functions will be used to evaluate greening measures under the common agricultural policy in different European regions.
Contact firstname.lastname@example.org for more information