The Problem

The risk posed by coastal flooding is often considered according to the relationship between the magnitude of flood events (best expressed as the elevation to which the water rises) and how often such events occur.  This relationship is termed a magnitude-frequency relationship (or recurrence interval) and is often expressed as a curve, such as in the example below.


These curves allow flood experts and coastal managers to determine the extent of defences required to provide to provide a certain level of protection. For example, if we wanted to build defences to protect against a 1 in 100 year event using the black line on the  curve above then we would require a defence that could withstand at least an 6.44 m high flood. The dashed line indicates how the magnitude frequency curve would change under a hypothetical 20cm of sea-level rise. For more on return periods/recurrence intervals see this post on SurgeWatch.

To construct magnitude-frequency curves requires data on historic floods through time. In the UK such data is provided by the the UK tide gauge network. The UK tide gauge network continuously monitors sea level around the UK. Much of the network was established following the North Sea disaster of 1953, though many records do extend as far back as the early 19th century. Combined with reliable Met Office hindcast data this gives coastal managers a reasonable understanding of storm surges in the last century. However, this data is usually only large scale and lacks the nuanced understanding of local impacts required to properly understand local flood risk. Further, the brevity of tide gauge datasets is itself a problem. Whilst an excellent source of high resolution, high quality data, tide gauge data only spans at most the last ~100y and significantly less in many other locations.

To create meaningful and high quality magnitude-frequency curves it is necessary to have data from as long a time frame as possible. The relatively short term nature of the tide gauge record is problematic in that; ​

  1. It raises into question the statistical validity of determining the likelihood of large events (1 in 100y or even 1 in 1000y) from such a short dataset​
  2. It may miss significant events associated meteorological processes which lie outside the range of the dataset

Thus it is necessary to attempt to supplement and extend the range of instrumental data with data from other sources, which is where this UK Coastal Floodstone Project and the wider project it forms part of come into play.