A Q&A with Oliver Wing: Fathom’s global reach

Articles 04.12.2017

Our Chief Research Office Dr Oliver Wing answers questions on global flood mapping

Prior to the onset of large-scale modeling, comprehensive flood maps were mostly confined to Western Europe, while America had lagged behind. How has Fathom overcome this?

Going back again to what has traditionally been done in hydrology, a field team of engineers would go to a specific catchment, collect flow data, survey the riverbed and the terrain, and then produce a model of that particular area. This was done at great expense and required a lot of manpower. To generate complete flood maps of, say, an entire country, this would be repeated in every catchment to form a nationwide mosaic.

This approach has been adequate for the more densely developed, wealthy countries of Western Europe, but is more problematic for the vast expanse of land in the USA with a sparser population density. As such, flood map coverage in the USA is very patchy as FEMA (the agency in charge of flood mapping) essentially had to triage their modelling strategy: smaller rivers and less developed catchments were often ignored.

As we use continental-scale data in order to build a flood model of the entire country, our approach is not subject to these coverage issues. Offering advantages in cost, time and ability to re-simulate, the Fathom approach seems to be the solution to the flood mapping problems modern governments are having to contend with.

How has Fathom achieved a system for global flood mapping?

Our methodology does not involve us being traditional hydrologists, where we would go to America to collect river flow readings and survey the topography. We sit in our offices and our algorithms churn through data that we acquire remotely. The most important aspect of a flood model is the terrain data. We have obtained this from a nationwide dataset which measures elevation derived from airborne lasers or satellite radars. River flow data is also extracted from existing stream gauges. Combining these two things with some clever statistics and interpolation forms the core of the modeling strategy, enabling us to simulate flooding everywhere.

You’ve worked with the likes of NASA. How have you gained successful global coverage?

Our work with NASA was borne of deciding to simulate forecasts, nowcasts and hindcasts of flooding resulting from the recent hurricanes in the USA to aid with the disaster response. Our work with other agencies (e.g. the Environmental Protection Agency and The Nature Conservancy) often arises from presenting our work at scientific conferences. We annually attend the autumn meeting of the American Geophysical Union, and it is there that scientists from those organisations encounter our products. It is usually from there that our commercial and academic relationships blossom.

Fathom has bridged the gap between academia and business while creating a commercially viable product – which can be a challenge. How have you managed that?

As we are all affiliated with the University of Bristol, we can tap in to the world-leading research that goes on there. The Hydrology Research Group has collaborated to create this suite of methods and they continue to build upon each others’ work. For instance, when Chris Sampson wrote the global modelling structure he used the base of the model written by Paul Bates in 2000, which was drastically improved by Jeff Neal in 2012. It all stems from this progression of talented scientists beavering away, which reached critical mass in 2013 when the company was formed, and Fathom realised that there was a gap in the market for precisely the kind of thing the research group was producing.

What are the next steps for Fathom?

Our recent peer-reviewed paper is the springboard from which we can progress. We’ve produced this product and we now have, in a peer-reviewed journal, shown that it is of use. From there, we can now target areas of the model identified in the paper as needing improvement – no model is perfect, or ever will be. We can also use the model to produce other things – for example Niall is building a stochastic model to simulate plausible events. At the moment our model just produces a return period flood extent everywhere, but that will not ever occur at one moment in time. Niall’s model will generate plausible events – simulating what could happen at one period of time, using the Fathom-US model as a base.