River floods cause some of the most destructive events experienced by societies (e.g. Kerala, 2018), with large numbers killed each years as a consequence, with flooding impacting on infrastructure, communities and economic development. Increasing exposure to flood events, through the settlement of unsuitable areas is changing the vulnerability of large populations globally.

The frequency of large flood events can be difficult to determine from short instrumental records, which are often less than 50-years in length, as such the accurate estimation of flood risk is challenging. However, communities and landscapes have long witnessed extreme floods and as a result have responded to these events in a variety of different ways over hundreds and thousands of years.

Communities have long recorded extreme flood events through historical accounts (old books, letters etc), in the form of flood marks or in other physical representations, such as paintings.  Past floods are also recorded within the landscape in both the sediments deposited (geological) and through the flood damage caused to living organisms such as trees (botanical).

Considering these different types of archives, provides a state of the art review, with a particular focus on recording mechanisms of flood information, the development of different methodological approaches and the type of information that those archives can provide. Past studies provide a wealthy dataset of hundreds of records of past floods, whose analysis reveals a noticeable dominance of past records in Europe.

This review in WIREs Water identifies how records of flooding in the past have, and continue, to improve flood hazard assessment, examining the potential they provide in areas where no instrumental records exist and how such data sources can be used in future flood management and mitigation planning.

Kindly contributed by Bruno Wilhelm.