top of page

USA

UK

USA

AUS

FP_710x1440 2.png

Real-Time Flood Forecasting, England

  • Alastair Sheppard
  • Apr 11
  • 3 min read

What our clients say:
The flooding mechanisms in the River Steeping catchment are complex due to the pumped sub-catchments and overall flatness of the catchment. However, the use of Flood Modeller allowed us to ensure that the floodplain and passage of floods through the catchment was accurately represented, whilst also allowing for hugely reduced run times, making the model suitable for flood forecasting.” 
Abbey Hewson, Senior Consultant, AECOM


AECOM was commissioned by the Environment Agency to develop at real-time flood forecasting model of the River Steeping, Lincolnshire, as part of the Modelling and Mapping (M&M) Framework. The model will be used by the Environment Agency for the timely and accurate provision of flood warnings to the public.


As part of the Environment Agency's M&M framework, a non-real time 1D-2D Flood Modeller-TUFLOW model had already been delivered by AECOM for the River Steeping catchment. Flood Modeller was used to convert the non-real-time model into a real-time 1D model suitable for flood forecasting.


The Steeping catchment is a complex catchment: gravity drained in the upper catchment and pumped in the lower catchment with heavily modified watercourses and linear flood defences. This meant that the most appropriate approach was hydrodynamic modelling, instead of using hydrological routing reaches, as more commonly used in flood forecasting.


Flood Modeller 1D flood forecasting model. Source: AECOM
Flood Modeller 1D flood forecasting model. Source: AECOM

As flood forecasting models are required to run quickly in order to predict when water levels or discharge will surpass a predefined threshold, the existing model was reverted to a simpler, 1D only model. The target was for the model to successfully run a 60-hour simulation in under 2 minutes. To achieve this, the following simplifications were made to the model:


  • truncation of the upper reaches of the model where no forecasting points were present;

  • stripping out cross sections to increase spacing to ~500m;

  • using extended 1D cross sections in the undefended floodplain/gravity drained catchment;

  • using 1D reservoir units to represent the various ‘compartments’ of the defended floodplain.


As a detailed mapping study had already been carried out, there was already a sound understanding of flooding mechanisms in the catchment. As such, 2D mapped outputs were used to inform the schematisation of the 1D reservoir units to effectively represent interconnectivity across the floodplain and overtopping of flood defences.




Key Facts


  • Conversion of a 1D-2D Flood Modeller-TUFLOW model to a 1D only Flood Modeller model for real-time flood forecasting

  • Hydrodynamic routing model

  • 99% reduction in run times achieved compared to the 1D-2D model

  • <5% difference in maximum water levels compared to the 1D-2D model




The model was run for a series of calibration and verification events as well as the 0.1% AEP design event. By comparing to the 2D mapped outputs, it was confirmed that the reservoir units representing the floodplain were filling and draining as expected and in the correct timeframe. This was important as correct identification of the timing of threshold crossings on the rising limb of the flood hydrograph are often more important than the prediction of the peak level itself.


Model calibration was also carried out for a tidal surge event. This determined that the model would perform even in the most extreme conditions. This is paramount as model failure would have a critical impact on the delivery of flood warnings to communities in the River Steeping catchment.


Overall, comparison of results with the non-real time model showed a <5% difference in maximum water levels. This demonstrated that the model results remained accurate despite the simplifications made. In addition, the model ran within target run times for all calibration and verification events, resulting in a 99% reduction in run time in comparison to the non-real time model for the 0.1% AEP event. Finally, the peaks and timing of threshold crossings matched observed data well, which is critically important for the timely and accurate provision of flood warnings.




Abbey Hewson


Abbey is a hydraulic modeller and hydrologist with experience across detailed flood zone mapping, flood forecasting, hydrological analysis for flood estimation and low flows, flood risk assessment and options assessment in the UK and the Middle East.

bottom of page