The authors are independent contributors to the Global Risk Institute. They are solely responsible for the content of the article.
This report summarizes the advances of the project since the last report submission. View Report 1
The overall goal of this study is to develop a framework for rapidly assessing the hazard (i.e. the probability and magnitude of a dam failure) and the exposure (what gets affected by a failure), scalable over many regions for a preliminary ranking of the priority areas of concern. The intended application is for a portfolio level risk analysis by investors, asset managers, and insurance providers.
In the first stage (2018), a climate risk model was developed to detect the atmospheric drivers of storms that could trigger a dam failure in the conterminous United States. Additionally, a dam risk score was proposed to rank individual dams according to the potential severity of damages caused by their failure, with and without considering the cascading effect of failure from upstream dams. The risk score could take values between 0 and 1.5 and was obtained with the elements described in Figure 1. This was exemplified with the analysis of 5,800 dams located in New York State. While this risk score can be a helpful first screening of the dams, its calculation had some important gaps. First, the inundation area in case of failure was not estimated using dam break analysis; it was subjectively defined with a buffer area around a downstream channel. The buffer extent was not linked to the characteristics of the dam (the same buffer area was used across dams) or the topography. Second, the element number 7 in Figure 1 (number of dams downstream that could be affected) was not based on analysis of the flow that could be released from an upstream dam and how this could affect it.
Therefore, the main focus of the past three months has been on improving the estimation of the exposure in case of failure. Existing tools and methodologies developed by US agencies were reviewed and tested using the Cumberland River Basin, a tributary of the Ohio River Basin as test case. We found that many programs were needed to perform the analysis (HEC-HMS, DSS-WISE, ResSim ArcGIS Pro, R, ResSim and Hazus), some of them were slow and unstable and required intense data manipulation. The results of the test case are provided in this report. In order to make this a rapid and scalable framework, the process needs to be streamlined. The focus of the following stage will be on a) the estimation of the probability of overtopping, and b) trying to make the exposure estimation more accurate and efficient.