The importance of input data quality and quantity in climate field reconstructions – results from the assimilation of various tree-ring collections

Franke, Jörg; Valler, Veronika; Brönnimann, Stefan; Neukom, Raphael; Jaume-Santero, Fernando

doi:https://doi.org/10.5194/cp-16-1061-2020

Articles | Volume 16, issue 3

https://doi.org/10.5194/cp-16-1061-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/cp-16-1061-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 16, issue 3

Research article

|

18 Jun 2020

Research article |

| 18 Jun 2020

The importance of input data quality and quantity in climate field reconstructions – results from the assimilation of various tree-ring collections

Jörg Franke, Veronika Valler, Stefan Brönnimann, Raphael Neukom, and Fernando Jaume-Santero

Download

Final revised paper (published on 18 Jun 2020)
Preprint (discussion started on 06 Aug 2019)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Review of cp-2019-80', Anonymous Referee #1, 11 Sep 2019
- AC1: 'Response to comments of reviewer #1', Jörg Franke, 02 Dec 2019
RC2: 'review Franke et al.', Anonymous Referee #2, 04 Nov 2019
- AC2: 'Response to comments of reviewer #2', Jörg Franke, 02 Dec 2019
RC3: 'Referee comments', Edward Cook, 06 Dec 2019
- AC3: 'Response to comments by Edward Cook', Jörg Franke, 19 Dec 2019

Peer-review completion

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

ED: Reconsider after major revisions (06 Jan 2020) by Keely Mills

AR by Svenja Lange on behalf of the Authors (09 Mar 2020) Author's response

ED: Referee Nomination & Report Request started (26 Mar 2020) by Keely Mills

RR by Anonymous Referee #1 (17 Apr 2020)

Suggestions for revision or reasons for rejection

I find the revised manuscript significantly improved over the original submission. I appreciate the efforts by the authors in addressing the issues raised by the reviewers. The description of the methods and presentation of the results are found to be clearer and more insightful. Some remaining minor issues are found in the revision, outlined below.

• Page 1, first line of abstract: Two very broadly defined factors in the quality of reconstructions are listed, methods and input data. Are there other factors? The use of “mainly” in the sentence seems to imply that other factors are at play. Perhaps delete the word?
• Page 2, line 40: “the switch from radiosonde to satellite observations”, I believe a more accurate statement would say “the addition of satellite to radiosonde observations”.
• Page 2, lines 52-53: I believe that your statement about some studies claiming that “higher quantity of input data would always be beneficial” is somewhat of a mischaracterization. Some of these studies do show that input of more data can have detrimental consequences on some of the characteristics of the reconstructions. Please revise.
• Page 3 line 109: “…fact the our prior”, “that” instead of “the”?
• Page 4, second line: “only one observation per year per record” to be more precise?
• Page 5, line 181: “little too much weight” is overly qualitative. This can still have some detrimental consequences and a statement reflecting that should be included in my opinion.
• Page 6, Table 1, line 4: “duplicate proxies cannot be excluded”: I believe they can, but in this experiment they are not. I would propose “duplicate proxies are not excluded” for a more accurate statement.
• Table 1, line 6, “got little weight”: is this clearly shown later or an hypothesis? As you explain later, relying on statistics can lead to mischaracterized error characteristics and weighting in the analysis. Please revise.
• Page 7, lines 220-223: If the majority of B14 records in the US do not contain information on temperature, and are appropriately weighted by the assimilation, why do we see an improved correlation, particularly over the eastern US, when B14 is used over PAGES records? If included but with little weight in the analysis, I would expect a neutral effect on skill over the prior. How do you reconcile? Please explain more clearly.
• Page 8, line 229 “Here, we find large differences”: Please indicate a corresponding figure to guide the reader.
• Page 8, line 231, “is again in the middle”: This is not clear. Please provide a more quantitative description of the comparison.
• Page 8, line 240, “correlation improvements”: Please indicate a corresponding figure to guide the reader.
• Page 8, your statement on lines 240-241: Just trying to summarize here: these regions of negative skill seem to be related to a combination of negative impact of PAGES data (Himalaya) (interestingly!), and B14 data forward modeled as temperature data only (India and US southwest), where N_TREND data do not influence positively. I would suggest that such details are worth including in the text.
• Page 8, line 242: Fig. 4d instead of Fig. 6d ?
• Page 8, lines 246-247, “precipitation skill clearly improves”: Looking closely at your results, I would add that this improvement is particularly noticeable over North America, even perhaps most over the eastern US. This is a finding consistent with results presented in Tardif et al 2019. The authors could consider mentioning this fact to further underline their finding.
• Page 8, sentence on lines 247-248: Are you referring to only precipitation here, or temperature as well. Please clarify.
• Page 8, line 255, “removes most of the negative RMSESS”: I feel this is a bit of an overstatement with respect to temperature. Negative RMSESS values persist over eastern North America for instance. The improvement seems more pronounced over Asia.
• Page 8, line 264, “due to overfitting”: And maybe due to duplicates? How can you distinguish? Please clarify.
• Page 9, line 272, “very few grid points with negative skill”: I again feel like this is a bit of an overstatement. I do not see changes in precipitation skill, some improvement in temperature, but negative RMSESS values remain over parts of North America (eastern Canada, southern US).
• Page 9, line 273: I find it awkward that results shown in Fig. 5 are not discussed before this point, and that only one frame of the figure is discussed in the entire paper.
• Page 9, lines 291-292: This is not a very clear statement. I would suggest that a more accurate one would read something like: Although not an issue addressed in this work, another study suggests that including the unscreened B14 records and modeling them using a similar approach than presented herein (including both temperature and moisture influences), can lead to problems in the representation of longer (than inter-annual) scales in temperature reconstructions.
• Page 10, lines 312-313: I am not sure I understand how increasing observation errors and its associated decrease in analysis skill can be interpreted as a sign of overfitting. Can you describe more clearly the link?
• Page 10, line 333: “underestimation of variability” ?
• Page 11, lines 355-356: I believe that Tardif et al. (2019) have shown a sensitivity as well. Maybe cite their work to support your statement.
• Page 11, line 367: “most probably”?

Hide

ED: Publish subject to minor revisions (review by editor) (21 Apr 2020) by Keely Mills

AR by Jörg Franke on behalf of the Authors (30 Apr 2020) Author's response Manuscript

ED: Publish as is (12 May 2020) by Keely Mills

Short summary

This study explores the influence of the input data choice on spatial climate reconstructions. We compare three tree-ring-based data sets which range from small sample size, small spatial coverage and strict screening for temperature sensitivity to the opposite. We achieve the best spatial reconstruction quality by combining all available input data but rejecting records with little and uncertain climatic information and considering moisture availability as an additional growth limitation.