This manuscript by Lyddon et al. clearly sets out an approach to identify tidal and river flood hazard thresholds for an estuary to determine the dominant or compound drivers of flooding. This understanding is important for local hazard forecasting and response planning. The methodology to identify and explore the impact of compound flooding uses national monitoring sources and freely available numerical models allowing application to other estuaries. The results presented demonstrated the application using a flashy catchment, representing the most vulnerable estuary setting in the UK. The results show the spatial variability in inundation area in response to different flood driver combinations, while the discussion provides insights to improve hazard warning and investigate future uncertainty in such catchment types.

The authors use of impact information (e.g., bus cancellations) is an innovative way to develop a historic record of flood events in the absence of documented records. The figures are clearly presented, and all are thoroughly discussed in the text to illustrate key messages/findings. The contour plots Figs 6-8 are a clear way to illustrate compound hazard impact and compare results, which could be applied to different studies to explore multiple hazard drivers.  

Minor queries/comments:

The hazard thresholds use sea level and river discharge. I assume the elevation and discharge choice was to reflect the long-term monitoring available, so the results can be used by authorities in future assessments using available data. A question for consideration in the reply to reviewers is: would there be any value in using two elevation metrics in the analysis?

In the abstract it should be clarified the time lag considered is 3hrs. Sensitivity to a range of time-lags is not considered.

When mentioning locations in the introduction (e.g., Lancaster and Humber) they should be located as NW or NE England.

L110, clarify if the record length is determined by the monitoring duration or the start of this study (using available data when the simulations were performed).

L130, 22:00 is pm the day before the peak river discharge. It’s not clear if the flooding occurred on the falling max tide or the incoming tide.   

Figure 2. Can Caesar-Lisflood provide information about the currents within the estuary to understand when slack tide occurs relative to the river flow? This would help understand the processes involved in holding the water within the estuary domain. The time lag is currently presented between Qmax and TWLmax. There could be asymmetries that prevent slack water occurring at TWLmax. It would be interesting to see the Qmax lag relative to slack tide (turning from flood to ebb). It would be worth checking the colour bar doesn’t constrain the lag so it is relative to the TWLmax tide. The compound flooding could occur on the next tide (more than +6hrs might represent flood tide again). If there are events that occur more than +/- 6hrs either side of TWLmax can they be identified (two panels maybe)?   

Figure 3 and 11. Using a different shape to explain the colour coding in legends (e.g., squares rather than circles) would help quickly separate this information from the circles used to indicate the top 50 Q.

L366, the flood extents aren’t clearly assigned to the results in Figure 9. How were the three scenarios in each subplot selected? I assume it was to illustrate a similar FloodArea as defined on L365. In Figure 9 the distance up estuary could be marked (or the place names added) to show the locations of sensitivity in Figure 10. To help show how the icon boxes (Fig 9) link to the results the grid could be plotted on the TWLmax and Qmax axis, with the grid boxes labelled as tide dominant, river dominant, compounded and moderately compounded. Putting the grid onto Figure 11, would help show the probability of occurrence of the different flood drivers.  

P20/21, check “FloodArea” is always in italics.

Figure 11, should P=0.9 in the caption be 0.6 to match the contours plotted? For the joint probability it needs to be clarified that co-concurrence means Qmax must occur in +/-6hrs of TWLmax (for a flashy estuary), they do not have to occur at exactly the same time.   

L684, have not.

Author contributions, should CN be NC?

The manuscript presents an interesting framework for setting thresholds for estuarine compound flooding using combined hydrodynamic statistical techniques. The approach is applied to Conwy Estuary, North Wales, a particularly vulnerable area to compound flooding. It represents a current thematic area and can be particularly useful in improving compound flooding assessment in estuaries worldwide. The study's main objective is to identify the coastal and fluvial conditions that lead to flooding in an estuarine system.

The manuscript is well-written (English level) and presents high-quality scientific content with interesting results and discussions. No doubt, a lot of work has been done, and the research is highly relevant. However, the connection between chapters (structure) needs improvement. In addition, the authors make some general assumptions throughout the text about specific terms that can confuse the reader, especially if he/she is not familiar with the United Kingdom or the field of coastal/estuarine flooding (especially in the Introduction and Methods Section). Further explanations need to be given in these specific points to improve the reader's comprehension for a broad scientific audience.