Effects of Bias-Correcting Climate Model Data on the Projection of Future Changes in High Flows

Round 1

Reviewer 1 Report

The main objective of the article is to quantify the efficacy of different bias correction methods on meteorological data at recreating the hydrological behavior in German catchments. The results would indicate which bias correction methods should be applied to climate data used in future hydrological scenarios which will inform water resource management. 

The article is well written, concise and does not try and be anything but an informative data paper. Very well done by the authors. 

The only thing I believe that is lacking, but important, is a quantification of the error based on the timeframe of selection for bias correction comparison (see Lafon et al., 2013 and Spellman et al., 2018 below). The timeframe selected to correct and derive and assess the metrics is 30 years which is a large window. If any change points in the mean and/or variance of the datasets occur (particularly for precipitation) and are not captured, this could have an impact on the conclusions and inferences drawn about he efficacy of each method. I suggest the authors test for whether or not there exist change points in the meteorological data and quantify whether or not these change points are captured in the observed and then simulated hydrological series. Then partition the dataset and conduct a sensitivity on whether or not deriving bias correction metrics for a set time frame have any impact on the overall output series within some reasonable threshold of error. The aforementioned suggestion would help improve the overall utility of the future simulated data on water resource management, as changes in the mean and variance of streamflow data are expected under future scenarios. Though these will inherently be captured by the increases and decreases in precipitation (and of course other variables) used to force the model, the error in potential changes that could occur within future simulated time frames can then be quantified. For more on how to approach this see:

Lafon, T., Dadson, S., Buys, G., & Prudhomme, C. (2013). Bias correction of daily precipitation simulated by a regional climate model: A comparison of methods. International Journal of Climatology, 33(6), 1367–1381.

Spellman, P., Webster, V. and Watkins, D., 2018. Bias correcting instantaneous peak flows generated using a continuous, semi‐distributed hydrologic model. Journal of Flood Risk Management, 11(4), p.e12342.

I would also include the equations for some of the bias correction methods. A little more information about their limitations and under what application and circumstances they are used would also be useful to readers. I was familiar with most methods, but had to do a double read on the ISI-MIP approach as that was a little harder to visualize. Again, a quick refresh would be nice as if someone is not familiar with all the methods, going back and forth between papers to obtain information may be cumbersome. 

Line 88: Need brackets. Where there is a random 2, should there be a [2]? This occurs throughout in other places as well....

Author Response

See the uploaded file.

Author Response File: Author Response.pdf

Reviewer 2 Report

The article “Effects of bias-correcting climate model data on the projection of future changes in high flows” evaluates the impacts of climate variability in high flows. I think this is an important issue due to the significant impacts of climate change in discharge.

The article is well organized, sometimes is difficult to follow because authors use many acronyms, nevertheless, the methodology is well explained. I think that although the bias-correction issue is discussed and analysed in many climate change impact studies some results of this work could be used in other mesoscale, flat, wet catchments. Nevertheless, I have some conflicts with the applied methodology that I think they should be discuss in the Discussion and Conclusions. As the authors say in the introduction section, “BC mostly impairs the alteration of the relations among climate variables and feedback mechanisms. The application of BC for the historical period as well as for a future period assumes that the bias is constant over time in a changing climate, which is not assured” (lines 63-66). However, this IMPORTANT fact is not analysed, not even mention, in the Discussion and Conclusions. They say “applying bias-correction on climate model data prior to the use as input data for hydrological models seems necessary in order to maintain the plausibility of simulated hydrological processes” (lines 411-412), considering that probably the future climate processes will change do the authors think that the bias correction of temperature, precipitation, wind, solar radiation… is necessary? I understand for precipitation but considering the less influence of the other climate variables in the hydrological modelling (figure 3) may be is better to don´t BC the humidity, wind…, what do you think? Continuing with the BC uncertainties, I think that you should mention that BC has more sense when the future climate is more close to the present conditions (better for 2020-2049 than for 2070-2099).

Finally, I would like to congratulate the authors because I think that you did a high effort using many climate scenarios, bias methods…

Author Response

See the uploaded file.

Author Response File: Author Response.pdf