Riverine flood risk screening with a simple network-based approach: A proof of concept in the Ganghes-Brahmaputra basin

Floods cause major problems around the world. Over 35 million people were affected by floods in 2018. They have a growing worldwide impact on life and property. Changes in climate conditions lead to unanticipated variations in glacial runoffs, snowmelt and precipitation, which all significantly changing river flows. An imbalance in river network equilibrium leads to flooding and often ends up causing tremendous damage to society and the environment. Regions that are perceived to be downstream from the source of flooding may end up taking the brunt of the river force due to flood cascades. Floods account for about a third of all natural catastrophes worldwide, they cause more than half of all fatalities and are responsible for a third of the overall economic loss.Modelling approaches are often used to determine flood consequences. Two types of flood models are commonly used: statistical models and flow simulation models. Statistical methods are easy to use but provide limited insight into flood problems. Flow simulation models’ results can be very accurate, especially for hydraulic simulation models. However, these models are expensive to use and develop, and they require a lot of data. These requirements make them unsuitable for application in developing countries and analysing large watersheds. Flood risk screening models try to solve these problems. They are suitable for use in data-sparse regions and are efficient in terms of omputational costs. However, there is a lack of knowledge between river structure and cascading flood effects, and there is a lack of models that are efficient, easy to understand, use topological data and have the purpose of risk screening. In this research, we show a flood model based on complex network theory to efficiently study the cascading effects of floods in riverine systems. Cascading effects are defined as floods that occur as a result of water waves through the system that originate from upstream sources. The developed model uses the hydrological Muskingum routing method. We found that it was possible, notwithstanding many assumptions and a lack of data, to reproduce system behaviour during an extreme flood event in the Ganges-Brahmaputra Basin. Satellite elevation data were used to construct the river network, and satellite precipitation data was used to feed the model. The model can indicate high risk reaches based on the simulated overflow, the flow exceeding a predefined capacity. No existing models are known that can do this, on a laptop, within seven min- utes per simulated day, with limited data for a watershed that exceeds the size of one million square kilometres. The network structure of the model makes it possible to achieve a better understanding between river typology and cascading flood effects. The model is not without its limitations. It cannot pinpoint when and where floods will occur, because it only calculates overflow. Moreover, flood failure mechanisms are not yet included in this model. Failure mechanisms will change model behaviour: when a flood occurs water temporarily leaves the system, which reduces downstream risk. Overflow cascades, therefore, would be shorter in reality than in this model. The model is a proof of concept that shows the potential of a network theory-based risk screening method in flood simulation context. Its properties make it suitable for analysing the effects of changing precipitation patterns, which, for example, could originate from climate change studies. Another use case is real-time forecasting of discharge levels if the mode is combined with real-time discharge levels and precipitations forecasts. The model can be used as an early warning system: alerting when and where high discharge levels are expected. We anticipate our model to be a starting point for policy screening and scenario analysis. Sugges- tions are made to include policy options within the model. Policy analysts can then use the model to compare different policy interventions for all kinds of (future) scenarios. The model should not be seen as a replacement of the advanced hydraulic simulation models, but as a complementary tool useful at an earlier moment in a design process with the purpose of screening options. Ultimately it can become a framework with the aim to support informed decision-making.https://github.com/bcvanmeurs/rnaEngineering and Policy Analysi