This technical report documents an initial, working implementation of a Bayesian framework for balancing BONSAI – a large hybrid Supply-Use Table (HSUTs), developed in the “Getting the Data Right” project. The aims are twofold: Firstly, to demonstrate feasibility and scalability. The current Stan implementation successfully balances HSUTs with more than 1.8 Million flows on a High Performance Cluster and produces posterior samples that can in principle be propagated to environmental footprints. The second aims of this report is to make assumptions and simplifications transparent. Several aspects of the model are deliberately simple (e.g. data-driven priors for flows, global uncertainty parameters for transfer coefficients and conversion factors, a basic set of soft constraints). These choices are documented, together with ideas for how they could be refined. The report and the results should therefore be read as a proof-of-concept and technical baseline, rather than a final, fully validated methodology.
Schulte et al. (Mon,) studied this question.