Technical note: Effects of uncertainties and number of data points on line fitting – a case study on new particle formation

Mikkonen, Santtu; Pitkänen, Mikko R. A.; Nieminen, Tuomo; Lipponen, Antti; Isokääntä, Sini; Arola, Antti; Lehtinen, Kari E. J.

doi:https://doi.org/10.5194/acp-19-12531-2019

Articles | Volume 19, issue 19

https://doi.org/10.5194/acp-19-12531-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/acp-19-12531-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 19, issue 19

Technical note

|

09 Oct 2019

Technical note |

| 09 Oct 2019

Technical note: Effects of uncertainties and number of data points on line fitting – a case study on new particle formation

Santtu Mikkonen, Mikko R. A. Pitkänen, Tuomo Nieminen, Antti Lipponen, Sini Isokääntä, Antti Arola, and Kari E. J. Lehtinen

Download

Final revised paper (published on 09 Oct 2019)
Supplement to the final revised paper
Preprint (discussion started on 11 Dec 2018)
Supplement to the preprint

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Review', Anonymous Referee #2, 29 Dec 2018
- AC1: 'Author's reply to reviewer 1', Santtu Mikkonen, 22 Mar 2019
RC2: 'Review of "Technical note: Effects of Uncertainties and Number of Data points on Inference from Data – a Case Study on New Particle Formation" by Mikkonen et al', Anonymous Referee #1, 29 Jan 2019
- AC2: 'Author's reply to reviewer', Santtu Mikkonen, 22 Mar 2019

Peer-review completion

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

AR by Santtu Mikkonen on behalf of the Authors (22 Mar 2019) Manuscript

ED: Referee Nomination & Report Request started (26 Mar 2019) by Fangqun Yu

RR by Anonymous Referee #1 (23 Apr 2019)

ED: Reconsider after major revisions (24 Apr 2019) by Fangqun Yu

AR by Santtu Mikkonen on behalf of the Authors (05 Jun 2019) Manuscript

ED: Referee Nomination & Report Request started (10 Jun 2019) by Fangqun Yu

RR by Anonymous Referee #3 (17 Jul 2019)

Suggestions for revision or reasons for rejection

I went through the 2nd revision of the manuscript and replies to the comments. I feel that this technical note provides some useful insights with regard to the uncertainties in line fitting of observational data, commonly used in deriving empirical relationship between new particle formation rates and precursor gas concentrations. However, I feel that several important issues raised by the previous reviewer were not adequately addressed. In particular, I think that the following issues should be addressed before the manuscript can be recommended for publication in ACP.

1. Data generation, Figures R1 & R2. The previous reviewer used about two pages to support his/her arguments with regard to the importance of the data generation procedure. I consider this to be an important and valid point. In contrast, the authors’ reply was too brief and not fully addressed the concerns raised. The authors should at least add a case to use the data generated according to the procedure suggested by the previous reviewer (evenly spaced rather than normal distribution). This may change the conclusion with regard to OLS, as demonstrated in Fig. R1 and Fig. R2.

2. With regard to using NPF data from CLOUD and Hyytiala, as original comments and replies copied below.
C: Since the authors likely have access to significant NPF data from CLOUD and Hyytiala, it is possible they could perform fits using what they believe to be the appropriate fitting method and compare them to the literature values. This would call attention to using the correct procedure, and would provide a database of corrected data for use by the aerosol community. This reviewer does not believe this is beyond the scope of this paper. It would involve a paragraph of introduction to the data, a description of the fit method selected, and a table of results.

A: It is true that we have access to these type of data but currently, as the data are not open, we do not have permission to use them in this study. We have made tests with data collected from Hyytiälä and San Pietro Capofiume, Italy and the results are similar than seen here with simulated data. However, we cannot publish these results and thus we will publish a Python tool for running some of the methods and encourage people to test that on their own data. The tool can be found in GitHub: https://gist.github.com/mikkopitkanen/da8c949571225e9c7093665c9803726e. The link is also added to the end of the manuscript.

I agree with the previous reviewer that it will be useful to perform fits to the real NPF data from CLOUD and Hyytiala and compare with the literature values. I feel that the authors’ explanation that they do not have permission to use them is not adequate. Firstly, many papers with regard to CLOUD and Hyytialla have already been published and these data should be available for others to use based on the data policy of many journals. Secondly, some CLOUD data are in the public domain (for example, see the supplement materials of Dunne et al., Science, 2016) and can definably be used. This manuscript will be enhanced and much more useful to the community if the authors “could perform fits using what they believe to be the appropriate fitting method and compare them to the literature values”.

3. First page of the reply, second line from the bottom. “We have added important information to text and supplement”. The previous reviewer asked for “This reviewer suggests a detailed re-write of section 3 to make clear the decisions made on the approaches used to generate the data.” I didn’t find substantial revision in Section 3. Also the supplement (S1) is totally on regression method, rather than on the data generation.

Hide

ED: Reconsider after major revisions (17 Jul 2019) by Fangqun Yu

AR by Santtu Mikkonen on behalf of the Authors (16 Aug 2019)

ED: Publish subject to minor revisions (review by editor) (02 Sep 2019) by Fangqun Yu

AR by Santtu Mikkonen on behalf of the Authors (05 Sep 2019) Author's response Manuscript

ED: Publish subject to technical corrections (09 Sep 2019) by Fangqun Yu

AR by Santtu Mikkonen on behalf of the Authors (09 Sep 2019) Manuscript

Download

Article (1438 KB)
Full-text XML

Short summary

Atmospheric measurement data never come without measurement error. Too often, these errors are neglected when researchers make inferences from their data. We applied multiple line-fitting methods to simulated data mimicking two central variables in aerosol research. Our results show that an ordinary least squares fit, typically used to describe relationships, underestimates the slope of the fit and that methods taking the measurement uncertainty into account performed significantly better.