This study provides a summary and comparison of the metrics used in the literature for rainfall event temporal loading and evaluates them for a large number of rainfall events in Denmark. The work is surely informative, but the question remains what new insights we can gain from it. Is there anything we do not know or cannot do now that this evaluation of metrics will enable us to do?  

In its current version, the analyses seems more suitable for submission in the form of a technical note. This would require drastically shortening the content, some suggestions and further comments are provided in the supplement.

The manuscript addresses a common issue when comparing different publications regarding rainfall events and their characteristics, namely that different authors and research groups use a wide range of metrics to characterize rainfall events. Therefore, this analysis, comparing and evaluating a range of metrics used in prior publications is a useful tool for researchers to interpret and compare results from various papers, as well as for authors and researchers to select meaningful metrics for their analysis in future work. As such this represents a valuable contribution to the field of rainfall event analysis.

However, the current manuscript contains certain limitations that require further analysis prior to publication. The decision to use 10 time points when calculating/representing dimensionless mass curves (DMCs) to compare the effect normalisation has on metrics seems flawed and represents a seemingly unnecessary own goal. Using 10 time points aggregates and summarizes the event time series, hence it is unsurprising that DMCs results for metrics that are sensitive to time discretization and aggregation show significant differences compared to the original time series data. Therefore, in its current state, the tests which are meant to examine whether normalization affects metrics seem to be flawed, as the method used both normalizes and aggregates the data. I strongly recommend that the effect of normalization be tested independently of any aggregation, by using the normalized, dimensionless mass curves at their original 5-minute resolution.

General comments:

The authors review and compare metrics for the “temporal loading” of rainfall events assessment i.e. the temporal distribution of the precipitation intensity within rainfall events. They carry out an extensive literature review and identify 52 metrics which have been applied to describe this temporal loading of events. They categorise these metrics into classification metrics, summary statistics and intermittency metrics. They apply the metrics for a large data set of rainfall events observed in Denmark and compare them regarding sensitivity to temporal resolution and redundancy. Finally, the recommend metrics for the characterisation of special aspects of temporal loading.

The authors have invested much time in analysing the large amount of literature and the many metrics. The paper is not innovative regarding new methods or better description of hydrologic phenomena. However, it is useful for the selection of appropriate metrics for own analyses, so it should be published after a revision.

It is somehow a mixture between a review paper, a research paper and a technical note. For a technical note it is too long for a pure review it contains too much calculations. Maybe it can be shortened to become a technical note. I am not sure if that is feasible but it would allow to concentrate on the most important results and the final recommendations. Also, it would avoid some lengthy descriptions which are partly tiresome to read. A few suggestions for shortening are given below.

Specific comments:

1. Introduction: The paper discusses temporal loadings of fixed storms based on the Eulerian view of the storms. The authors should mention, that both Euler (fixed storm) and Lagrange view (moving storm e.g. described using radar data) would be possible to calculate such metrics. In addition, the metrics analysed purely describe 1D temporal loadings. The natural storms however have space-time dimensions, so 2D temporal loadings may also be possible.

2. Literature review: The details how the literature review has been carried out could be omitted if the paper is to be shortened.

3. Statistics regarding the literature review e.g. table 1 or fig. 3 are not really necessary if the paper is to be shortened.

4. Recommendations: This should be one of the main outcomes. The equations for the recommended metrics should be given.

5. References: The references should be listed beginning with the last name of the first author.