Reconstruction of Hydrometeorological Data Using Dendrochronology and Machine Learning Approaches to Bias-Correct Climate Models in Northern Tien Shan, Kyrgyzstan

Round 1

Reviewer 1 Report

This manuscript is on very good studies on relevant subject that has applications.

The introduction is too long and verbose. It should be restructured to bring it to 60% of the present.

Discussion should also be rewritten in order correlate results obtained and the importance of the work done.

Conclusions are vaguely presented and voluminous. Thorough revision of the manuscript should be done and the language of the text may be evaluated by and expert.

I have tried to rewrite the abstract of manuscript of paper (please enclosure).

The manuscript of the papers is not acceptable in the present form.

Comments for author File: Comments.pdf

Author Response

Please see our uploaded file in which we respond to all comments by Reviewer #1.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Editor.

I have finished my review on the proposed paper “Reconstruction of Hydrometeorological Data Using Dendro-chronology and Machine Learning Approaches to Bias-Correct Climate Models in Northern Tien Shan, Kyrgyzstan” water-1798649-peer-review-v1.

 

Summary of the manuscript:

In the proposed paper, the authors’ goal is to reconstructed hydrometeorological time series using dendrochronological data. They collected 33 increment cores from Picea abies (L.) Karst and created the dendrochronology time series from 1886 to 2013. Then, using temperature, precipitation and discharge records from 1915 to 2013, they manage to reconstruct the respective climate time series.  

 

General review:

1. Generally, the manuscript presents a very interesting topic and the specific research seems to include some significant points for the research community of this field.

2. The proposed paper is very well written with very good use of English language. Except some very minor grammatical mistakes and word errors. The authors should check again the paper to correct these minor mistakes.

3. The proposed paper is very well structured. It begins with an analytical Introduction with the appropriate references that helps the reader to get into the subject immediately. In Introduction there is an effort to provide previous studies with similar scientific content, which took place in the research area and in other countries. Authors describe and set very well the scientific problem and how other researchers have approached. At the end of Introduction, authors clearly state the goals of the research.

4. The methodology is generally very interesting, and well explained, so other researchers could easily repeat it. However, there are some issues that need clarification (see below comments).

5. The results scientifically explained and are OK, except some points (see below comments).

6. The quality of the work in Discussion is generally ok.

7. Conclusions are appropriate for this paper.

 

Additional points for revision:

In my opinion, the proposed paper could be characterized as a good research work, complies with aims of WATER. 

Page 2: “Tree rings can provide…. instrumental record”. This statement needs support with literature. I proposed to add the following two studies: (Kastridis et al. 2022, Vitali et al. 2018).

3. Results: Where are the results of the dendrochronology analysis? It is supposed that the reconstruction of the climate data is based on the created dendrochronology time series. This analysis is very significant and showing the quality of the selected cores and the reliability of the dendrochronology time series. You should add a subsection with this analysis adding, a figure with the non-standardized and the standardized annual tree-ring indexes, the sample depth, and the standard statistics such as RBAR, EPS, SNR, AC1 and MS. The sample depth is very crucial especially for the years before 1900. For example, how many trees are dated from the year 1886?

Figure 2: The R² has no meaning is charts with time series (years in X axis). The R² is useful in scatter plots depicting observed and modeled values. I order to see if there is significant trend in time series, you can calculate the MANN-Kendal trend test.

4. Discussion: This part is weak. More literature should be added and comparisons with other similar studies in other locations should be done.

In tables 4 and 5 you should add the Standard Deviation (SD) of the observed values. Is the only way to interpret the RMSE values, if they are acceptable.

 

References

Kastridis, A.; Kamperidou, V.; Stathis, D. Dendroclimatological Analysis of Fir (A. borisii-regis) in Greece in the frame of Climate Change Investigation. Forests 2022, 13, 879. https://doi.org/10.3390/f13060879.

Vitali, V.; Büntgen, U.; Bauhus, J. Seasonality matters—The effects of past and projected seasonal climate change on the growth of native and exotic conifer species in Central Europe. Dendrochronologia 2018, 48, 1–9. https://doi.org/10.1016/j.dendro.2018.01.001.

Author Response

Please see the attached file where we respond to Reviewer 2's comments and suggestions.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Dear authors.

Thank you very much for the provided responses to my comments. I believe that the paper has significantly improved. However, I have some final comments that should be addressed in the paper. 

1. The tree sample the year before the 1914 are very few. As i can see from the graph the trees before 1914 are 5. As you understand the period of reconstruction depends on these 5 trees, a fact that significantly increase the error and the reliability of the resulting reconstructed time series. You should definitely address this in the paper as limitation.

2. You gave the values of RMSE, for example in table 4 and 5. RMSE describes the difference between model simulations and observations in the units of the variable. RMSE optimal value is zero (0). However, the statement “optimal value is close to zero” is relative and the RMSE alone is not very informative. For example, when the RMSE of a hydrological model in a river (with discharge values ranging between 200-600 m³/sec) is 2.292, could be considered very low and close to zero. However, in another case, for example temperature simulation model (with temperatures values ranging between 0.3 and 5.7 ^oC), an RMSE of 2.292 is huge.

So, the best way to interpret the RMSE and understand if is acceptable for model evaluation, is to provide the Standard Deviation (SD) of the observed data. It is known from previous studies that RMSE values less than half of the SD (Standard Deviation) of the observed data may be considered low and acceptable for model evaluation (Singh et al. 2005, https://doi.org/10.1111/j.1752-1688.2005.tb03740.x.). The ratio SD/RMSE should be below 0.70. You provide the SD in tables 4 and 5, but as I can see the values in green cells are barely acceptable. Discuss in the text the above mentioned, and explain if the RMSE is acceptable in terms of the SD. Add the proposed and other literature to support your discussion.

3. In introduction (lines 33-35) you address the potential problems associated with the climate change. Here, you can rephrase the sentence adding the problems of biotic and abiotic dangers (insect outbreaks, wild fires) that are in increase due to the climate change. Here with [2-5] references, I propose to add the following: Kastridis et al. 2022, https://doi.org/10.3390/land11060911. For example: "Moreover, the frequent occurrence of extreme weather and climate events such as heatwaves, droughts, heavy rainfalls and biotic/abiotic catastrophes during recent years are the evidence of climate change [2–5], which is linked to global warming".  

 

 

 

 

 

 

Author Response

Please see attac

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

This paper has certain scientific value, but the reliability of its method and conclusion is still questionable and needs to be improved. 

Comments:

1. There are many modes in CMIP5 and CMIP6 projects, and only a few modes are selected in this paper. Then, what is the basis of selection and why are these modes selected? The reason needs to be stated in the article.

 

2. The first line in Table 2, the two Pa and Tmean_a, is confusing as to what they stand for.

 

3. In Fig. 2, why only the data in the early period is reconstructed, but not the data in the later period? There is almost no overlap between reconstructed data and observed data, so how can we convince the effect of reconstructed data?

 

4. The results of CMIP5 and CMIP6 are more representative of the decadal scale climate change, and their spatial resolution is very rough. The observation site in this study is located in a mountainous area with complex terrain and great altitude variation, where precipitation and temperature have strong spatial heterogeneity. The reliability of the methodology and the scientific nature of the conclusions drawn from direct use of global climate models compared with observations from two or three river valley sites is questionable. At the very least, climate simulation results need to be downscaled to fully consider the difference between the altitude in the model and the altitude of the observation site, and the model results need to be revised before comparison.

 

5. It is not sufficient to use only tables and Taylor diagrams to represent the comparison between the results of models and observation sites. It is suggested to increase the presentation and comparison of time series changes (such as interannual scale). 

Reviewer 2 Report

- My main criticism is about increasing the temperature and discharge. How the authors can justify this? We expect by increasing the temperature, the evaporation increased, too, which affects overland flow. 

- Another main issue is about the performance of the GCMs models.  According to the results, the CORDEX REMO models indicated the greatest performance (KGE=0.24). But a perfect value for Kling-Gupta efficiency (KGE) is equal to 1. Thus, the models can not simulate the observations well. 

Reviewer 3 Report

The objective of this study is to show the importance of GCM spatial resolution and new parameterization of physical processes in a region of complex orography using historical simulations from the Climate Model Intercomparison Project phase 5 (CMIP5), CMIP6, and Coordinated Regional Climate Downscaling Experiment (CORDEX). This manuscript is well organized and the drawn conclusions are coherent with the obtained results. However, references should be updated to include more recent studies.

Lines 27 – 28: To arrange the keywords alphabetically.

Line 59: It is Multi-Model Ensembles (MMEs)

Lines 65 – 66: I think that you should add these recent references to support your sentence “Selected GCMs should replicate the observed climate in terms of spatial and temporal variability”. I would like to suggest:

Smeraldo, S., et al. (2021). Generalists yet different: Distributional responses to climate change may vary in opportunistic bat species sharing similar ecological traits. Mammal Review, 51(4), 571-584.

Du, Z., et al. (2021). Potential geographical distribution and habitat shift of the genus Ammopiptanthus in China under current and future climate change based on the MaxEnt model. Journal of Arid Environments, 184, 104328.

Line 86: To delete the double dots.

Line 98: It is Bias-Corrected (BC).

Line 115: Please, add the north symbol and the scale in figure 1a.

Line 150: Please, move this table in the supplementary material.

Lines 317 – 318: To delete (Figure 2a).

 

Line 349: To delete (Table 5).