I really enjoyed reviewing this paper, it provides an excellent examination on the value and importance of using historical records to improve precision and accuracy in flood frequency analysis, particularly for the largest flood events. The authors present a compelling case for the utility of historically augmented records within flood frequency analysis. I felt the case study was particularly valuable as it demonstrates how such information can be embedded within and used to support and question conventional flood risk assessments. The authors cover all the key areas, including non-stationarity of the series, whilst providing a detailed case study which adds to the growing evidence base demonstrating the importance of historical flood records in FFA globally.

I have provided an annotated copy of the manuscript with comments, minor amendments and thoughts, I hope that the authors find this helpful in making their revisions.

My only more substantive comment relates to some of the image quality, whilst the figure content is fine, several of the figures may need to be presented at a higher resolution (e.g. Fig 4), I do though recognise that this may simply be a presentation issue with the review manuscript file size.

Neil Macdonald 

University of Liverpool

The article presents a probabilistic model for flood frequency analysis that uses the number of times a perception threshold is exceeded over a historical period and considers the uncertainty of discharges during the systematic period. The experiments are interesting and well planned. A key novelty of the model is to recognise the imperfect nature of both the perception threshold and the length of the historical period by making them parameters of the probabilistic model.

The five model variations are clearly explained; however, their differences could be made clearer by presenting these in a table so that readers can refer to them more easily. Uncertainty in the results is considered throughout, which is a strength of the article.

The results section is lengthy with 13 figures, several of these are multi-plot. The paper would benefit from choosing the key results to present and focussing the discussion on these.

The conclusion summarises the main findings well but could include more recommendations for use of the model in practice and linking back to the wider picture of flooding.

The structure of the article is clear and generally well written. Please see the supplementary attachment for minor comments.