Parametric Flood modelling for the vulnerable population in Laos

Parametric solutions can close the insurance gap while providing a protection for many developing nations in the world experiencing losses from natural catastrophes. The process generally involves real-time data analysis of environmental variables to verify the intensity of the event and if it passes or not the threshold specified in the policy. Despite the different benefits, developing effective policies is still very challenging due to regional variations in the parameters and the scarcity of high-quality and accessible data. This applies particularly to floods, which are also very difficult to model accurately given their complex nature. 

In this study it is shown how GallagherRe has faced these challenges in developing a flood solution for the vulnerable population in Laos. The workflow developed relies on the Mekong River Commission river water level gauge data, which is assessed for quality in reconstructing selected historical events. To evaluate their intensity, a Generalized Pareto Distribution is fitted to the statistically independent extreme values extracted from the data for return period estimation, enhanced through Monte Carlo simulation. The information is then used to identify an equivalent flood extent derived from third party hazard maps for the catchments assigned to the selected gauge stations through an event agnostic approach. The reconstructed extent is finally intersected with the input risk to get an estimate of the vulnerable population affected. 

The quality control of the gauge stations data identified that, due to a change in water level regime caused by anthropogenic events such as upstream dam regulation, only 16 out of the 28 available gauges can be used to support the parametric scheme. The limited catchments coverage determined for the valid gauges still allowed a significant portion of the risk to be captured in the hazard maps. In fact, most of the selected events resulted to be driven by the main Mekong River and its major tributaries, areas with both good valid gauge coverage and high population density. Despite this gap, it is also shown a positive correlation of increase in estimation with increasing size of event.