Spatio-temporal modelling of lightning climatologies for complex terrain

Simon, Thorsten; Umlauf, Nikolaus; Zeileis, Achim; Mayr, Georg J.; Schulz, Wolfgang; Diendorfer, Gerhard

doi:https://doi.org/10.5194/nhess-17-305-2017

Articles | Volume 17, issue 3

https://doi.org/10.5194/nhess-17-305-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

https://doi.org/10.5194/nhess-17-305-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 17, issue 3

Research article

|

06 Mar 2017

Research article |

| 06 Mar 2017

Spatio-temporal modelling of lightning climatologies for complex terrain

Thorsten Simon, Nikolaus Umlauf, Achim Zeileis, Georg J. Mayr, Wolfgang Schulz, and Gerhard Diendorfer

Download

Final revised paper (published on 06 Mar 2017)
Preprint (discussion started on 04 Jul 2016)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review of Simon et al. paper titled "Spatio-temporal smoothing of lightning climatologies"', Anonymous Referee #1, 01 Aug 2016
- AC1: 'Response to RC1', Thorsten Simon, 08 Aug 2016
RC2: 'Review of Simon et al.: 'Spatio-temporal smoothing of lightning climatologies'', Anonymous Referee #2, 11 Aug 2016
- AC2: 'Response to RC2', Thorsten Simon, 23 Aug 2016

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

ED: Reconsider after major revisions (further review by Editor and Referees) (02 Sep 2016) by Andreas Günther

AR by Thorsten Simon on behalf of the Authors (19 Sep 2016) Author's response Manuscript

ED: Referee Nomination & Report Request started (27 Sep 2016) by Andreas Günther

RR by Anonymous Referee #1 (04 Oct 2016)

Suggestions for revision or reasons for rejection

I have read the revised paper and find that the paper is somewhat improved, but I still have some issues with the methodology and results.

1. The analysis of the lightning data using the GAM is simply a fancy way of smoothing the data spatially and temporally. Such smoothing functions (whether splines or interpolations) are available in most software packages (Matlab, IDL, R, Python, etc.) and hence this is not a new development of the authors. And I still wonder why the authors think their GAM is better than simply smoothing the raw data.

I agree that the observed data is noisy, but maybe for a reason. Maybe the noisy topography in the region results in peaks of lightning above mountain peaks, less in valleys, and when the model smooths the data these maxima disappear. Smoother data is not always better or closer to reality. Maybe the noisier real data is better for determining risks.

2. The authors use 3 parameters only to describe the lightning climatologies (altitude, day of year, and longitude/latitude). However, in reality there may be many other parameters that determine the lightning distribution. For example, vegetation cover, slope of topography, soil moisture, etc. Hence, the model can only be as good as the input paramters.

At the bottom of P5, the authors state that the model is generated with 5 years of data, and then tested against the 6th year of data. However, I do not see these comparisons. Where is the predicted distribution for year XXXX next to the observed (smoothed) distribution for year XXXX. Such comparisons are necessary to show that the three parameters used are sufficient for building the climatology.

Minor comments:

P1 line 12: Simply smooting the observed data would also produce a climatology that varies smoothly over space and time. This is not unique to your method.

P2 line 21-23: Why is using the raw data a problem for quantitative assessments? The same method can be followed to assess the risk using the raw data, however noisy is may be. You can use simple spline or interpolation to smooth the observed data and get similar results.

P3 line 19: Why was only Carinthia used in this study, and not the whole of Austria. I guess the data is available, so why not use it?

P3 line 29: Is this the number of flashes over the 4 month period? Please clarify.

P6 line 11: Fig.4 is still not clear regarding the y-axis units. Are these the weighting functions? Are they dimensionless? What is the physical meaning of 0.5? What is the physical meaning of a negative value? Please explain. If it is not clear for me, I guess others will have difficulty as well.

P6 line 22: We don't need the model (or the data) to tell us that lightning is mainly in the summer in Austria. I think this is well known.

P6 line 28: If we already know all of this from previous studies, do we need another paper to make this statement? It is difficult for me to figure out what is NEW in this analysis.

P9 line 21: The prediction tool will not fall below the climatology only if all relevant parameters are included in your model. As mentioned above you only used 3 parameters. And we have not seen any comparison between lighning predicted using your model for say, 2015, compared with the observed climatology for 2015. Please present such a comparison, including the correlation coefficient between the predicted and observed distributions.

Hide

RR by Anonymous Referee #2 (12 Oct 2016)

ED: Reconsider after major revisions (further review by Editor and Referees) (27 Oct 2016) by Andreas Günther

AR by Thorsten Simon on behalf of the Authors (16 Jan 2017) Author's response Manuscript

ED: Referee Nomination & Report Request started (18 Jan 2017) by Andreas Günther

ED: Publish subject to technical corrections (08 Feb 2017) by Andreas Günther

AR by Thorsten Simon on behalf of the Authors (13 Feb 2017) Manuscript

Short summary

The study presents a newly developed statistical method to assess the risk of thunderstorms in complex terrain. Observations of lightning serve as an indicator for thunderstorms. The application of the method is illustrated for Carinthia which is located in Austria, Europe.