Reacción a "Reactions: Study blames climate change and other human factors for severe flooding in Libya and Greece"

In my view, the Daniel episode that affected Greece and then Libya may have been of greater intensity than would be expected with the historical climate because of the large volume of precipitable water now in the atmosphere, largely related to the huge thermal anomaly in the oceans. And with climate change, the oceans will undoubtedly continue to warm much more. Therefore, such intensities are more likely to recur in the very near future. The formation of Mediterranean cyclones is not a new phenomenon, and even assuming the possible doubts about whether they will form more frequently, what we can clearly state is that, once they do form, the intensity is directly related to the amount of precipitable water, which is increasing. 

As far as the research is concerned, the first thing I must emphasise is the speed with which this study, which is so complicated from a climatological point of view, was carried out. The study is of more than sufficient quality to be taken into consideration. However, some minor comments are:  

1. The limitations of ERA5 for such extreme cases need to be analysed.  
2. Figures 7 and 8 do not look good, but it seems that in the first case there is some statistical relationship between temperature and maximum rainfall in 4 days for each year; not so in Figure 8, at least not so clearly.  

The study period probably needs to be further extended to at least 100 years of historical data (observed or simulated) for the case of Libya. Mediterranean climate has so much variability that it is very difficult to find statistically significant trends in a few decades. 

It is very difficult to measure the probability of such extreme events that have a frequency of less than "once in 100 years", as there are hardly any observed data to test and validate the models. Any record of previously unobserved hourly intensities is by definition an increase of "1 now versus 0 before". Therefore, it is necessary to use theoretical functions to extrapolate a probability of phenomena that have never occurred. The technique of using theoretical distributions is very difficult to verify for variables such as precipitation as they are far away from typical Gaussian behaviour, as is the case for temperature.

I think the authors should also use observed data from long weather stations (at least 50 or 100 years) to better analyse the return periods. Models and reanalyses are insufficient for these cases of extreme rainfall events [as in the case of those occurring in Spain]. In fact, the parameterisation of the models is, in general, more suited to typical rather than extreme rainfall situations, and therefore tends to underestimate sub-daily intensities.