A cloud-based end-to-end Flood Early Warning System for Greater Wellington

<p>Te Pane Matua Taiao – Greater Wellington Regional Council (GWRC) in New Zealand is a local government organisation responsible for flood risk management along 320 km of river channels and 280 km of flood banks. Recent flood exposure assessments involving a regional scale flood model identified almost 200,000 people, a third of the region’s population, are exposed to flooding, a number expected to increase by around 16% over the next 100 years in response to climate change.</p> <p>Despite large exposure, no major flooding has occurred in Greater Wellington over the past 2 decades. This resulted in a lower prioritisation of flood forecasting capabilities with out-dated operational models only available in longer response time catchments. By leveraging an open distributed hydrological model concept, Wflow, and a configurable Delft-FEWS (Delft Flood Early Warning System) Greater Wellington has updated its flood forecasting capabilities and extended lead times of warnings.</p> <p>Delft-FEWS uses a combination of observation data, three local WRF models with different Numerical Weather Prediction products (ECMWF, UKMO and NCEP) and a Quantitative Precipitation Estimate (QPE) to generate hydrological forecasts for the Hutt River catchment. Hydrological model development was done using the global-to-local HydroMT model builder, where by leveraging open global datasets an initial model can be configured in reduced time compared with traditional methods. The model is iteratively updated by using a local data catalogue of standardised inputs, which can also be reused for any catchment in the region as a starting point for model development.</p> <p>Common limitations for implementing flood forecasting systems are IT infrastructure and hardware capabilities of organizations. Over the past years many organisations have migrated operational flood forecasting services to the cloud, which has several advantages, such as scalability of computing power, accessibility, back-up and redundancy options and improved security over traditional forecasting systems. Delft-FEWS is cloud compliant and was implemented in an Azure cloud environment hosted by GWRC.</p> <p>By integrating data, models and forecasting practices in a single flood early warning system, Greater Wellington Regional Council will have improved access to hydrological forecasts, improved forecasting capabilities and can act with longer lead times. River basins in Greater Wellington are small and steep with short response times ranging from 1 hour to 12 hours, also highlighting the need for an automated end-to-end forecasting approach. These short lead times have made accurate flood forecasts a challenge in the past, but the confluence of higher resolution, gauge corrected and QPE rainfall forecasts, institutional development, as well as the integration in an operational Flood Early Warning System, GWRC will be able to provide better and more actionable flood forecasts in the future.</p>