Use Bayesian belief network models to integrate diverse social and natural science models of cost-effectiveness of nutrient abatement in the Western Vansjø catchment
i) Reduce uncertainty in network models of effectiveness of fertilisation and ploughing measures by including additional process and calibration against new monitoring data
ii) Reduce uncertainty in nutrient abatement cost models by adequately accounting for opportunity costs of switching production processes.
iii) Farmer response: conduct a farm-level survey to obtain primary data on farm production functions for main crops in Western Vansjø sub-catchment and nonagronomical factors affecting abatement measure implementation.
iv) Develop model of farmer response to abatement measures based on opportunity costs and qualitative non-agronomical factors
v) Adapt the Bayesian network to MCMC simulation results from SWAT and MyLake.
vi) Assess the effectiveness of abatement measures in terms of changes in predicted water use suitability by adapting the results from a choice experiment study on water quality suitability thresholds.
vii) Use Bayesian network to integrate (ii-vi). Assess 1) joint uncertainty of cost-effectiveness of abatement measures, 2) the role of implementation probability in providing more realistic assessments of abatement measure cost-effectiveness under uncertainty and 3) the role of drivers in inflating uncertainty due to multi-correlation.
viii) Interact with managers and farmers in explaining the results of the Bayesian network.