Grant, G.R.; Keller, E.D.; Christophersen, A.; Mourot, P. 2023 A Bayesian Network pilot study to assess the probability of inundation of the Kauaeranga Spillway, Thames, New Zealand. Lower Hutt, N.Z.: GNS Science. GNS Science report 2023/06. 34 p.; doi: 10.21420/SQ5K-NN88
Abstract
Coastal inundation is likely to pose an increasing risk to people and infrastructure in many regions of Aotearoa-New Zealand with changing climate. Here, we investigate the use of Bayesian Networks (BNs) as a statistical and graphical approach to emergency flood management in the Hauraki Plains, New Zealand. Specifically, BNs offer a complementary and efficient alternative to the use of multiple computationally intensive process-based models (e.g. vegetation, land use, soil, groundwater, fluvial hydrology, coastal hydrology) when assessing the likelihood of inundation and/or flooding events. The Waikato Regional Council (WRC) identified the operation of the Kauaeranga Spillway, which forces closure of State Highway 25 and the main access road in and out of Thames, as a key uncertainty in their regional emergency management. The WRC guided the development of the research question and have an interest in using the model output. The question selected for this pilot study “What is the likelihood that the Kauaeranga Spillway will be in operation [road inundated] in the next six hours?”. In this report we outline the process undertaken by GNS Science in this pilot study, including engaging with the stakeholder, developing a proof-of-concept model and interpreting the results. We conclude with a reflection of what worked well, work required to validate the model and what needs improvement to apply this approach in future to other questions and/or situations. (auth)
