Abstract

Reed, D.W. 2008a. Eight things you don’t want to know about flood risk. Proc. Flood and Coastal Management 2008 Conf., Manchester, 1-3 July 2008, 8pp.

  1. You can test a model of most things. But you will never validate a flood-risk estimate. If you are careless or unlucky, someone may invalidate your estimate.
  2. If you are interested in estimating extremes you need to study information over a long period. You will find all sorts of reason not to.
  3. Met Office archives hold very extensive datasets. Only about 10% of daily rainfall data from the period 1881-1960 have been computerised.
  4. There is uncertainty about uncertainty.
  5. Unless the effect on the flow regime is very marked (e.g. reservoir construction, urbanisation), it is generally infeasible to demonstrate that a particular land-use change is responsible for trend in floods at the river-basin scale.
  6. Soils and geology are hugely influential. Packages rarely shut themselves down with the helpful message: Do not use me on this catchment. Stop and think!
  7. The toughest estimation problems are those where flooding is influenced by two or more variables. In principle, such problems can be dealt with by joint probability methods. It is well known that dependence between input variables can swell extremes of the output variable. A less discussed feature is that sequencing and storage effects are often influential in fluvial-fluvial, fluvial-pluvial, fluvial-tidal, pluvial-tidal and surface-groundwater problems. Storage provides flexibility but adds complexity. A common thread is that, where storage plays an important role, conditions experienced in most years may be a poor indicator of the conditions in exceptional events.
  8. Weaknesses in the river modelling approach to flood-risk mapping are discussed.
  9. Beware the unexpected. Major interventions change flood regimes. Where storage is introduced, the system will become sensitive to longer storms. The location of storage is hugely important. Storage on a minor headwater remote from the main flood-risk site is worth little.