Modelling Flood Risk at Global Scales: 130 million km2 at 30m Resolution in Just 3 Months
Traditionally, flood risk mapping studies are conducted at the scale of individual river reaches a few 10s of km in length, however important classes of flood risk management decision require analyses at much larger scales. These include insurance and reinsurance pricing, investment portfolio and supply chain analyses and determining an evidence base for government spending decisions. To satisfy these needs, researchers have increasingly sought to develop automated methods to build hydrodynamic models over large areas to undertake risk mapping at scale. In turn, this has required numerous developments in global terrain and hydrography data sets, methods for determining extreme flows, rainfalls and coastal water levels anywhere on the planet and new data sets for large area model validation. Latest global methods can simulate the entire 130 million km2 terrestrial land surface (excluding Antarctica and Greenland) at 30m spatial resolution in just a few months on just 2500 cores by using hyper-efficient numerical schemes and code. This talk charts the improvement in skill of such methods over the last decade and asks how much further we can go in producing global methods that have local skill.
