Review Reports

Round 1

Reviewer 1 Report

This manuscript is solid overall, and I recommend publication with only minor modifications. It does present data from a single station which limits its general interest, but that is outweighed by the unusually long record from the station. Also, Toronto's location is interesting in that it is in the transitional zone where winter temperature ranges are such that rainfall and snowfall are both common. One would expect that rain/snow ratios should show high sensitivity to the warming trend.

The paper is well written and the research methods are solid with one exception, the treatment of autocorrelation. The authors are correct to acknowledge that it is a potential problem, and the Durbin-Watson test is the appropriate one to use. Once it tests positive for autocorrelation, however, they should not cite statistics (particularly p values) that are based on the assumption of no autocorrelation they have just shown is invalid. Furthermore, using the non-parametric Kendall's Tau is not appropriate, either. It just relaxes the normal distribution assumption but still needs all pairs to be equally likely which is not true due to the lack of independence caused by presence of autocorrelation.

A preferable approach would be to fit the trend and a low order autoregressive model simultaneously using maximum likelihood. The does add a little complexity, but not much as an AR1 (or AR2 at the most) would probably be sufficient. There are many common software tools that will do this type of modeling, and the time series is plenty long enough to support estimating another parameter or two. Since the normal distribution assumption appear not to be a problem, this approach would produce statistically justifiable parametric estimates of the magnitude and significance of the trends.

Author Response

Please see attached

Author Response File: Author Response.docx

Reviewer 2 Report

 

Thank you for providing me the opportunity to review the paper by Gough and Hewer on winter precipitation patterns in Toronto.  I enjoyed the piece and I found the paper to be well organized and somewhat well written. The authors should be reminded that “data” is the plural of “datum” and should be treated accordingly throughout the manuscript (no more “data was”)!  The first sentence of the abstract seems grammatically incorrect (169 years was analyzed??).  Otherwise, the key references are cited, the abstract is informative, the length is fine, the methods are appropriate and innovative, the graphics are publishable, and the results are interesting.

Author Response

Please see attached.

Author Response File: Author Response.docx

Reviewer 3 Report

I enjoyed reading this manuscript and found it to be very interesting. The authors have made use of a unique long-term dataset to assess winter season precipitation phase. In general, the presentation of the material was good. The methods were appropriate and both the methods and results were well described. The conclusions presented are supported by the results.

My comments are primarily related to spelling and grammar. I have found a few minor issues that I outline below, but I encourage the authors to proofread again.

Line 10 (whole abstract): Check spacing throughout. There doesn’t appear to be a space at the end of each sentence but this could simply be because I am viewing a PDF that was probably converted from a Word document.

Line 19: I believe the authors mean “downtown” not “down”.

Line 126-127: On days with mixed precipitation the total precipitation was equal to the sum of rainfall and snowfall. It may be worth specifying here again that this would be snow water equivalent.

Line 130: The sentence reads "trace (0.2mm) of greater" and I believe the authors meant "or" rather than "of".

Line 150: Can you please explain how the very snowy or very rainy categories were calculated for the PPI? The snowy/rainy categories are described but the very snowy/very rainy are not.

Line 310: Presumably the authors mean "panels" not "panes".

Line 417: I think the finding that the SD:PD ratio was the best performing metric is interesting, especially given that most studies use SF:TP.

Author Response

Please see attached.

Author Response File: Author Response.docx