Flood Memory


Changes in the frequency and severity of floods is under close scrutiny due to increased storminess in projections of future climate. The Flood Memory project examines observed records of storms to help us better understand how clustering of hydrological events may obscure or even exacerbate climate induced changes.   The project investigates the most critical flood scenarios caused by sequences or clusters of extreme weather events striking vulnerable systems of flood defences, urban areas, communities and businesses.  The team analyse and simulate situations where a second flood may strike before coastal or river defences have been reinstated after damage, or householders and small businesses are in a vulnerable condition recovering from the first flood. By examining such events and identifying the worst case scenarios, we hope our findings will lead to enhanced flood resilience and better allocation of resources for protection and recovery.


The current practice of modelling flood risk based on single event simulations for fixed river cross-sections is incorrect, as it fails to consider active changes to channel shape via erosion and deposition processes. As these sediment transport processes occur both during a flood “event” and in the intervening low flow “recovery” period, there appears benefit in developing a coupled flow-sediment modelling approach to flood risk assessments which includes longer-term multiple event simulations.  The Heriot-Watt team is lead by Dr. Heather Haynes with specific objectives that include:

  • To establish the significance of flood sequencing (i.e. event clustering, recovery duration) on sediment transport regime, channel morphology and floodwater conveyance
  • To determine the sensitivity of flood risk assessments to changes in channel morphology
  • To ascertain the reciprocal relationship between flood defence asset and sediment transport


  • Based on a time-series of river flow gauge data, a hidden Markov modelling methodology has been developed to realistically re-order the sequence of flow events (100 scenarios).
  • Coupled 1D flow-sediment transport models within the Eden catchment, Carlisle, have been run continually (50 years) using different flow sequences.
  • The sensitivity of channel geometry and floodwater conveyance to flow sequence is sufficient to compromise standard freeboard allowance of flood defence schemes.

Related Publications

  • Pender, D., Patidar, S., Pender, G. & Haynes, H. (2015) Stochastic simulation of daily streamflow sequences for UK rivers using a hidden Markov modelling approach. Hydrology Research.
  • Pender, D., Patidar, S., Pender, G. & Haynes, H. (accepted) Incorporating sediment-related uncertainty into 1D modelling of fluvial flood risk. Journal of Hydraulic Engineering.
  • Pender, D., Patidar, S., Pender, G. & Haynes, H. (2015) Incorporating river bed level changes into flood risk modelling. IAHR Congress, Netherlands 2015.
  • Pender, D., Shvidchenko, A., Haynes, H. & Pender, G. (2014) The influence of flood sequencing on the morphology and bed composition of the American River, California, USA. IAHR River Flow, Lausanne, Switzerland 2014.