Deltares Nederhoff, Discusses Software To Project Flooding Impact
BY STAS MARGARONIS
Kees Nederhoff is a coastal modeler at DeltaresUSA with 5 years of experience in developing and applying Deltares software products. Before joining DeltaresUSA, he was employed at Deltares in Delft, The Netherlands and worked on a range of national and international projects including numerous World Bank projects, the USGS-Deltares co-operation and the Dutch flood safety assessment. Kees acts as a liaison between Deltares and the United States Geological Survey (USGS) in Santa Cruz and with San Francisco Estuary Institute (SFEI) in Richmond.
Kees says “the approach to storms and flooding in the Netherlands is to anticipate the problem before it happens and invest now rather than pay later. Even so, the country prioritizes what areas to invest and protect and what to emphasize less. Also, in the Netherlands, there is more acceptance of the need to spend money proactively (and thus higher taxes) but also to support good roads, bridges, ports, airports and rail service.”
In the United States, “we see the trend in which the investment is not made until the damage is done and rebuilding necessary. The cost of rebuilding is usually much higher than the cost of prevention.”
For example, Google is building a new headquarters in Mountain View which is protected from flooding by a levee that keeps water out from the Bay. This is a $1 billion investment. The cost of replacing the levee is likely to be much less.
Deltares is an independent institute for applied research in the field of water and subsurface. In the U.S., Deltares is mainly known for its development of computer models (e.g. Delft3D). The software that Deltares has developed is based “on experience with coastal flooding that we have seen in the Netherlands and have transposed to projects for the United States and other foreign users.”
SFINCS is “Deltares’ new reduced physics model that can compute overland flooding +/- 100 times faster than traditional models. SFINCS stands for Super Fast Inundation Solver for Coastal Systems. With a model that can compute flooding much faster it is possible to take into account combination of drivers of floodings.”
Kees says that in the current approach of computing the 1/10 year food plain, flooding is “based on only one of the drivers of flooding (e.g. river discharge). However, recent hurricane impacts (e.g. Irma or Harvey) have shown that flooding is often compounded by flooding due to rivers, local rainfall and coastal processes. Subsequently, the 1/100 year flood plain could be much larger than currently expected. For example, Hurricane Harvey damaged 204,000 homes of which three-fourths were outside of the 100-year flood plain. Those homeowners did not have flood insurance . The challenge is that we all must make sure we are getting all we can from past weather events so we are prepared for the future.”
Machine-driven algorithms are often put forward as the ‘solution’: “However, one of the main limitations with machine-driven data algorithms is that it only computes what it has seen before. Therefore, you still need real historical data to base your computations on. The challenge is to better extract the existing data and this is where AI and machine learning might help. However, there is no magic wand here. Hard work by engineers to compute what we have not yet seen based on computer models that compute how the water might flow and including different drivers of flooding, is.”