Transdisciplinary knowledge integration and embracing of uncertainty with Bayesian Belief Networks in water management

Climate change alters the water cycle, which makes adaptation in water management necessary. The interdisciplinarily coordinated transdisciplinary project KlimaRhön aimed at developing adaptation strategies in water management in the UNESCO biosphere reserve Rhön in Central Germany. Experts agree that scientific and stakeholders’ knowledge should be involved to develop adaptation strategies, which requires good integration tools.

To identify adaptation strategies in the participatory process of the project KlimaRhön, we integrated the knowledge of hydrologists, sociologists and stakeholders while embracing uncertainty with a Bayesian Belief Network. For this, sociologists introduced that the acceptance of relevant actors is needed to implement adaptation measures and hydrologists introduced a range of potential future changes in groundwater recharge. In the first workshops of the participatory process, the stakeholders jointly identified possible adaptation measures. In the last workshops, we focused on two adaptation measures, which are 1) the protection of springs taking into account pasture water supply and 2) the fusion of small water suppliers in a region, and planned to find strategies to stimulate their implementation. For this, the stakeholders identified the relevant actors whose acceptance to implement those adaptation measures is needed. Then, the stakeholders identified and weighted factors that influence the acceptance of the relevant actors to implement the respective adaptation measure. This knowledge was then integrated in two Bayesian Belief Networks (for two adaptation measures) and a suitable communication of the Bayesian Belief Networks, which also focused on the communication of the embraced uncertainty, was developed. In the last workshop in the participatory process, the Bayesian Belief Networks were presented to the stakeholders and discussed. In the presentation, the stakeholders could explore which combinations of factors can enhance the acceptance of the relevant actors for the adaptation measure and thus the probability that they implement it for different degrees of climate change impacts.

The conditional probability tables for the Bayesian Networks were derived directly from the stakeholder weightings. Thus, stakeholders did not need to fill out conditional probability tables, which would have been difficult for most of them, and time-consuming. Bayesian Belief Networks show the uncertainty of possible future conditions through the many possible combinations of factors, which might have enhanced the understanding of stakeholders for the need of flexible adaptation strategies. The stakeholders appreciated the good overview of the many interdependencies and their influence on the acceptance of the relevant actors to implement the adaptation measure. In this contribution, we present our integration approach, the Bayesian Belief Networks, its communication as well as its evaluation by the stakeholders.