Why research the history of flooding? Well, if one suspects that the catchment has no history of flooding prior to recent land-use changes, it is good to confirm this before arguing too hard that land-use change or climate change is to blame.
But what if the catchment has a forgotten history of flooding? Is it credible that the river system has undergone so much change that past vulnerability is no longer relevant?
Of course, it's easier not to bother researching the flood history, and get on with some interesting modelling. If the site floods later, you are safe in the knowledge that flood estimation is inherently uncertain. You can divert attention onto the wide variation between estimates by different methods ("These hydrologists can't make their mind up") or lay the blame on good-old climate change.
But don't be surprised if someone bites back by researching the local flood history. If the dice have rolled in such a way that the site has many large floods in its history, but none lately, it will be all too obvious that your design was based on a gross underestimate.
It's not all bad news. Knowledge that the site was untouched in an exceptional flood will confirm that the site is much less prone to flooding than others near the river.
Historical flood review isn't just for anoraks. It really is useful.
Get your dates right. A faulty date can lead to much fruitless searching.
Value contemporary accounts. Local newspapers weren't always given to sensation and opinion more than factual reporting. In their heyday, titles such as the Halifax Courier and the Inverness Courier were peerless in conveying all kinds of news to their readership. The Times is of special significance, both for its authority and because it is indexed from 1790. Newspaper reports often refer to earlier floods. Look out for rainfall data (not just general weather information), photographs of inundated areas and structural damage, and readers' letters in subsequent issues.
Local libraries are great, but the British Library National Newspaper Library can also be pretty useful.
Value contemporary maps (e.g.). They help to indicate the extent to which the river system has (or hasn't) changed in the last 150 years. There was an era when land liable to flooding was often explicitly marked.
Value daily rainfall data held in the National Meteorological Archives in Exeter and Edinburgh. In some cases, you can study the data cards filled in by the observer, and find additional notes about exceptional events.
Consult the BHS Chronology of British Hydrological Events. This is principally dedicated to events earlier than 1930. Treat the more anecdotal entries with a pinch of salt, but use them as a prompt to search for harder information.
Consult the British Rainfall yearbooks. They are most useful for events between about 1860 and 1970, and are excellent for helping to get flood-dates right.
Watch out for under-statement in wartime.
Do web searches. I now recognise that my forecast about the growing searchability of archive material from the 18th and 19th centuries (p. 56 of FEH Vol. 1) was too optimistic. With many tens of kilometres of shelved material in a typical County Record Offices, the proportion that is searchable electronically remains very small. Nevertheless, it's amazing what is out there, just a couple of clicks away.
Look at source documents where possible. Learn to distinguish original and derivative accounts, placing greater weight on the former.
Download a copy of CEH guidance on the use of historical flood data in flood frequency estimation. But don't be too hooked on a particular way of using historical data. Knowing how many exceedances there have been of a locally meaningful high threshold can be enough to indicate whether the current flood estimate is reasonable.
Finally, urge Government to make meteorological data efficiently and freely available to all: to practitioners not just to academics.