Flood prediction applications and techniques

Published on by in Business

We are preparing a multi-disciplinary project for industrial-academic research in the UK. The quality of flood prediction is quite poor - geographic granularity can be no less than tens or hundreds of metres. Also, advance warning on flood risk due to surface water flow from unregulated regions is predicated on real-time data, which excludes satellites. So, as we complete our proposal preps I'm looking for evidence or working concepts, applications, models etc which are in use and are providing as close-to-real-time prediction and advance warning as possible.

Taxonomy

6 Answers

  1. Hello,

    I think in terms of model approach, for some type of rapid flooding such as flash flood, a lumped hydrological model coupled with rainfall-runoff model would be one of the best approach. The reason is that with very small lead time, a more sophisticated approach such as hydrodynamic model will consume a lot of time to run. You can take a look at BROOK90 or GLOBAL BROOK90 lumped hydrological model as an example. One of the latest development is that it uses global climate data taken from satellite as well as global circulation model to cover area with less spatial data. Another example is that in Saxony, Germany, where I did some modeling related to those lumped model, there is also “extruso” framework which provide the regional model as well with higher resolution of spatial data.

    On the other hand, for a long lead time flood event such as plain flood, a 2D hydrodynamic model is sufficient to provide inundation extent, etc. Various software is available such as Mike11, HEC-RAS, Delft 3D, etc.

    1 Comment

  2. Catastrophes in Brazil happen yearly. Nevertheless we know where, how and why they happen. We do have anticipated catastrophes due to unlikely reasons as seismological events in an area which such events are not considered. Other reasons are also known and predictable. Nevertheless, nothing is done to mitigate those risks. As an example, I am attaching a research done by the Minas Gerais federal university study, showing seismological events and the tragedy in Mariana city in 2015. The study includes many papers and anticipated the tragedy of Brumadinho in 2019.

    1 Comment

    1. Thanks Silvio. I'm busy right now but I will share more detail on the approach we are looking at.

  3. Satellite based data may be integrated into a complex rainfall runoff model for real time prediction of flood magnitudes. The use of historical data as input to the models have challenges of data gaps.

    1 Comment

    1. Thanks John. We're looking at this approach, but the frequency of EO data places to much emphasis on the modelling for my liking as surface topology can change between events...

  4. It is just not the radar data that is important but, ground truthing of various parameters like, geology, hydrology and soil structure become very vital in predicting the flood warning.  We may have consider the dam structures around as well as they would pressurise the normal conditions.  I further feel the time of flood warning - could be different in different seasons as well.  Therefore, one has to compile historical data and match it with the ground truth for a better understanding.

    Answered on by

    1 Comment

    1. Agreed, and this is why we are intending to take a systems-based approach to the project as the end-to-end interaction of system elements is vital to the accuracy of the prediction achieved.

  5. Take a look at Coastal Risk Consulting (who, despite the name also cover fluvial and pluvial flooding).  They produce Lidar-based flood risk assessments for individual properties which take into account the variations in freeboard around each property, local drainage patterns and so on that can mean that two adjacent houses actually have different flood risks.  

    I would also argue that there is a time as well as a spatial dimension here.  There is a concept to be had of "virtual elevation".  If you have a house 10 feet above flood level and a 6 ft levee, you may be said to have 16 ft of virtual elevation.  If you have the same house with flood pumps instead of a levee, that can move 3 feet of water, you have 13 ft of virtual elevation.  The point being that virtual elevation can change day by day - the pumps might fail or gophers might undermine the levee - in which case your elevation is back to 10ft.  If, as well, you have a blocked storm drain or a building site pushing water onto your property, you may be below 10 feet.  This is not an abstract risk - virtual elevation changes on a given day or month could easily mean that the access to say a hospital emergency rom is under 18 inches of water that no-one had predicted ahead of time.