Towards operational impact based flood forecasting in Norway

Impact based forecasting is identified as a major challenge for hydrometeorological services by WMO. Impact based flood forecasting aims at closing the gap from forecasting how high the streamflow might be in the coming days to forecasting what the high flows might do. The target audience for impact-based flood forecasts can be both the population at large and the authorities responsible for civil protection and emergency management at national, regional, and local levels. Since 1989 The Norwegian Water Resources and Energy Directorate (NVE) has been running the national services for forecasting floods and issuing warnings for regions and municipalities. Precipitation-runoff models for almost 160 catchments distributed all over Norway are used to forecast streamflow. Flood warnings are then mainly issued based on predefined thresholds for streamflow (mean flood, 5- and 50 years floods). The warnings mention typical impacts of the expected floods but not specific impacts. NVE aims to provide impact-based flood forecasts for Norway and has started a 4-year pilot for four selected catchments in Norway to assess the information, models and tools needed to achieve this aim. This presentation will present the approaches selected and discuss their challenges. The model chain used to assess impacts needs to represent the relevant processes. In a first step we focus on riverine floodings. A first challenge is that the hydrological models need to provide forecasts where the flood might have an impact, which signifies that the hydrological models will be applied in ungauged locations. Subsequently, hydraulic models can be used to estimate water levels and depths in susceptible areas. We evaluate an approach where the models are configured to sub-reaches of the rivers were the up- and downstream boundary conditions are well defined and forecasted streamflow at selected points is used as input. The hydrologic model can be used to establish an archive of flood levels and or depths to reduce computing time. This approach is challenging when downstream boundary conditions are dynamic, e.g., close to sea with tidal influence or inland lakes where the water level peaks later than discharge in the upstream rivers. The water levels and depths are subsequently used to estimate impacts based on the buildings and infrastructure potentially affected by the flood. Here we can use the type of building (storage, residential, school etc) type of road (municipality, regional or national road) to assess impact. An important challenge here is how to weight and aggregate potential impacts.