A 148-year precipitation oxygen isoscape for China generated based on data fusion and bias correction of iGCMs simulations

Chen, Jiacheng; Chen, Jie; Zhang, Xunchang John; Peng, Peiyi; Risi, Camille

doi:https://doi.org/10.5194/essd-2021-460

Preprints

https://doi.org/10.5194/essd-2021-460

Preprints

01 Apr 2022

Status: this discussion paper is a preprint. It has been under review for the journal Earth System Science Data (ESSD). The manuscript was not accepted for further review after discussion.

A 148-year precipitation oxygen isoscape for China generated based on data fusion and bias correction of iGCMs simulations

Jiacheng Chen¹, Jie Chen^1,2, Xunchang John Zhang³, Peiyi Peng⁴, and Camille Risi⁵ Jiacheng Chen et al.

Show author details

¹State Key Laboratory of Water Resources & Hydropower Engineering Science, Wuhan University, Wuhan, 430072, China
²Hubei Key Laboratory of Water System Science for Sponge City Construction, Wuhan University, Wuhan, 430072, China
³USDA-ARS, Grazinglands Research Laboratory, 7207W. Cheyenne St., El Reno, OK 73036, USA
⁴Chongqing Southwest Research Institute for Water Transport Engineering, Chongqing Jiaotong University, Chongqing, 400016, China
⁵Laboratoire de Meteorologie Dynamique, IPSL, CNRS, Ecole Normale Superieure, Sorbonne Universite, PSL Research University, Paris, France

Received: 21 Dec 2021 – Discussion started: 01 Apr 2022

Abstract. The precipitation oxygen isotopic composition is a useful environmental tracer for climatic and hydrological studies. However, the observed precipitation oxygen is limited at both temporal and spatial scales. Isotope-equipped general circulation models (iGCMs) can compensate for the temporal and spatial discontinuity of observation networks, but they suffer from coarse spatial resolutions and systematic biases. The objective of this study is to build a high-resolution precipitation oxygen isoscape in China for a period of 148 years by integrating observed and iGCMs-simulated precipitation oxygen isotope composition (δ¹⁸O_p) using data fusion and bias correction methods. The temporal and spatial resolutions are month and 50–60 km for the isoscape, respectively. Prior to building the oxygen isoscape, the performance of two bias correction methods (BCMs) and three data fusion methods (DFMs) is compared after post-processing of eight iGCM simulations. Results show that the outputs of the Convolutional Neural Networks (CNN) fusion method exhibit the strongest correlation with observations with correlation coefficient mostly larger than 0.8, and the smallest bias with root mean square error mostly smaller than 2 ‰. The other two DFMs also perform slightly better than the two BCMs, which show similar performance. Thus, precipitation oxygen isoscape is generated by using the CNN fusion method for the 1969–2007 period in which all iGCMs have output and by using the bias correction methods for the remaining years. Based on the precipitation oxygen isoscape, the spatiotemporal patterns of δ¹⁸O_p across China are investigated. The generated isoscape shows similar spatial and temporal distribution characteristics to observations. In general, the distribution pattern of δ¹⁸O_p is consistent with the temperature effect in northern China, and with the precipitation amount effect in southern China, and be more specific in the Qinghai-Tibet Plateau of China. The δ¹⁸O_p time series mirrors a fluctuating upward trend of the temperature or precipitation in most regions of China. The temporal and spatial distribution characteristics of the generated isoscape are consistent with the characteristics of atmospheric circulation and climate change, indicating successful assimilation and extension of the observed precipitation oxygen isotopes in time and space. Overall, the built isoscape is reliable and useful for providing strong support for tracking atmospheric and hydrological processes. The dataset is available in Zenodo at https://doi.org/10.5281/zenodo.5703811 (Chen et al., 2021).

Jiacheng Chen et al.

Status: closed

RC1:
'Comment on essd-2021-460', Liheng Wang, 12 Apr 2022
The stable isotope composition of water is a very useful tracer to elucidate its formation history and cycle. The precipitation is an essential input for the hydrological cycle but it is very difficult to obtain continuous temporal and spatial variations of precipitation isotopes. This paper provides a high-resolution precipitation oxygen distribution in China based on IAEA long-term data and other datasets. These results will be of significant interest to the scientific community, particularly in hydrology. Overall, this review paper is scientifically robust given the available datasets. It is also reasonably written and well organized. I think the paper suits the reader of this journal and can be accepted for publication. I have some major comments and several minor comments for the authors to consider.

Major:

The topic of this paper is very interesting, in my opinion, it upgrades the data from a sparse point scale to a continuous regional area scale. I suggest that authors highlight this in the introduction section, and be more explicit about the meaning of the paper, which will help arouse the reader's interest and facilitate its further spread.

As we know, isotopic composition in precipitation also varies dramatically over time. Therefore, I think the temporal resolution is also important. What do you think about this? I think authors should state or discuss clearly the time resolution.

Minor:

Line 25: investigated —> shown?

Line 40 add reference: IAEA/WMO (2022). Global Network of Isotopes in Precipitation. The GNIP Database. Accessible at: https://nucleus.iaea.org/wiser

Line 63 watershed —> catchment

Figure 1 the unit (m) should be added to the legend.

IAEA also provided long-term data in Haikou, why isn’t it being used in your paper?

Line 96 There are many types of averages, are you using a monthly precipitation amount weighted average here?

Line 102 What is ï¼ would you like to give more information?

Line 219: Figure 3. I think the NW result is bad too. It should be discussed in detail.

Figure 5 – 8 What are SPR, SUM, AUT, and WIN? Jan- Mar is SPR? Authors should clearly indicate the months covered by each season. Table 2 also should be clearly mentioned.

Line 315-319. Author should add some references to prove your point.

Line338-40: You discussed TP again, why not move to Line302?

Line346: It's better to start a new paragraph so that it can be read more clearly.

Line 352-354: It's hard to see a significant trend.

Figure 9 Need to add y-axis labels.

Figure 10 I think it's very unclear to discuss why it is divided into these stages according to Figure 10. Based on methods? Or data? Authors should state it clearly.
Citation: https://doi.org/10.5194/essd-2021-460-RC1
- AC1: 'Reply on RC1', Jiacheng Chen, 18 Apr 2022
  
  We thank the reviewer for the valuable comments. We have responded to each point of the review in the attached supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-460-AC1
CC1:
'Comment on essd-2021-460', Nancy Swift, 26 Apr 2022
The authors create a high-resolution precipitation oxygen isoscape dataset for China by fusing eight iGCMs simulations and in-situ observations based on data fusion and bias correction techniques. I appreciate the authors’ great efforts to develop such a dataset, but I am quite dubious about the general content, and have large concerns about the novelty and quality of the dataset.
My main concerns are:
Data-quality and novelty:

It feels like a direct comparison of five commonly used fusion methods for developing a high-resolution dataset in China, without any new advanced fusion methods. Furthermore, the quality of the developed dataset is still questionable and unreliable due to its poor and insufficient present form. It seems to me that the intended novelty might be a high-resolution dataset.
The methods used to develop the dataset are unclear and not robust. For example, the sensitivity of model parameters are not evaluated and discussed; thus, the results are not robust. Why does CNN perform much better? Why is it set to a three-layer structure model? The observed data covers a short period and is not sufficient to train the model.
Using the interpretations of spatial pattern at the seasonally averaged scale (1969-2007) and the temporal pattern at the regional scale to validate the effectiveness and reliability of the data is not persuasive. Why not give us a comprehensive assessment at finer scale, such as time-series comparisons between the gridded simulations (50km) and in-situ observations at each station, and spatial patterns for each month. Without these comprehensive evaluations at finer scale, I am quite dubious about the data-quality and usefulness of this data set.
The presentation of dataset.

Introduction: The section of Introduction is not well written, and lacks to interconnect of the data it shares to and to show how it is valuable in relation to the Earth’s system. For example, readers should have a clear understanding of the motivation of this study, the purpose of creating such a dataset, which can be seen from a literature review of precipitation oxygen isoscape in hydrological and biogeochemical cycles. Then, followed by a detailed description of the available datasets, we could not find any description of the previously evaluated performance of iGCMs simulation in the Introduction. A description of the data-fusion method should also be added.
Data set and study area: Please add more details about the in-situ data (e.g., time-series of available in-situ data, number of data points at each station) in the supplementary material to justify the machine learning. For better visualization, a mesh of the iCGMs could be added in Fig.1
Methodology: This section is not well written, and should be clarified. For example, did you correct for iGCMs simulation bias at grid scale (grid by grid) or regional scale (using all gauges and all iGCMs within a specific region). How do these methods generate 50 km simulations from various iGCMs with different spatial resolutions?
Results and discussion: see above comments. Many published ESSD papers have demonstrated the uncertainties, limitations, advantages and 1-3 specific applications (for validation, evaluation, and analysis) of their dataset, which are also highly recommended for your paper.

For above reasons, I do not support its publication in the ESSD, without advanced approach or comprehensive evaluations of dataset at finer scale. Anyway, I still look forward to seeing the reviewers’ comments and editor’s decision.
Citation: https://doi.org/10.5194/essd-2021-460-CC1
- AC2: 'Reply on CC1', Jiacheng Chen, 31 May 2022
  
  We thank the reviewer for the valuable comments. We have responded to each point of the review in the attached supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-460-AC2
CC2:
'Comment on essd-2021-460', Donal Logue, 28 Apr 2022

General comments:
Obtaining a high-resolution long-term precipitation oxygen isoscape dataset can be critical for relevant hydrological studies. This study presented a first attempt to solve this issue, really appreciate, but from my point of view, it is still a rather premature dataset and have limited value. The authors downscaled and fused eight iGCMs precipitation oxygen isoscape using five different methods from a coarse spatial resolution (~2/3 degree) to a higher spatial resolution (0.5 degree). However, the work is not innovative and no important and robust findings were obtained. The results do not convince me since this method highly depends on the training data, which cover a short period, are unevenly distributed across regions, and insufficient to train model. Moreover, as we know, the precipitation oxygen isoscape is highly dependent on local climate conditions, terrain factors, as well as large-scale atmospheric and local circulation. No such physical-based ancillary data were used in this study, which limited the further applications of produced 0.5-degree data. The sub-pixel spatial patterns within a coarse pixel changes, but the current methodology cannot get this information. For the implementation of models, uncertainty of the datasets resulting from the model structures, parameters, training and testing strategies is not even discussed. Data-quality assessment at finer scale is poor presented. I have concerns about the reliability of spatial-temporal variations in your new data product at fine resolution.
In conclusion, this is a good attempt to generate a high-resolution dataset based on the fusion of in-situ data and iGCMs simulations, however, due to the above-mentioned issues, I don’t think this dataset meet the high standard of ESSD.

Citation: https://doi.org/10.5194/essd-2021-460-CC2
- AC3: 'Reply on CC2', Jiacheng Chen, 31 May 2022
  
  We thank the reviewer for the valuable comments. We have responded to each point of the review in the attached supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-460-AC3
RC2:
'Comment on essd-2021-460', Anonymous Referee #2, 14 May 2022
As one of the key tracers of hydroclimate change, precipitation isotopes have very important research significance. According to the current situation of less observational data, the authors used iGCMs to obtain high-resolution precipitation isotope data over the past 148 years, which will provide important data support for the study of precipitation isotopes and hydroclimate change. I think this research is meaningful, but there are still some problems in the article, I think it needs to be further elaborated before accepting publication.

There is a problem with regional division. For example, Yunnan Province is a typical southwestern region of China, and its climate is dominantly influenced by the Indian summer monsoon, which is different from the region of southeastern China where is mainly influenced the East Asian summer monsoon. How can it be divided into southeastern regions? Generally, the Indian summer monsoon precipitation oxygen isotope values during JJAS are lower than the East Asian summer monsoon rainfall. I don’t think current regional division is scientific.

There are some GNIP stations with long-term precipitation isotope monitoring data, which can be used to compare long-term changes with the simulation results, such as the Hong Kong station. I suggest that the authors can compare the Hong Kong station and other stations with 8-10 years of monitoring data with the simulated isotope records. This comparison can be used to verify the reliability of the simulation.

At present, the altitudes of each monitoring site vary greatly. When the author uses the monitoring data of each site for spatial analysis and compares them with the simulation results, did the author consider the effect of altitude on precipitation isotopes to calibrate? For example, according to the relationship between altitude and precipitation isotopes at each site or the large region scale, first calibrate the precipitation isotope data of all stations to the same altitude, and then compare it with the simulated data. Because the simulation precipitation isotope results are at the same altitude, if no calibration is performed, the spatial comparison of the simulated and monitored precipitation isotope data will inevitably not be the real results. In the current manuscript, it seems that the author has not calibrated, and it is suggested that the author add relevant correction processes or solutions.
Citation: https://doi.org/10.5194/essd-2021-460-RC2
- AC4: 'Reply on RC2', Jiacheng Chen, 31 May 2022
  
  We thank the reviewer for the valuable comments. We have responded to each point of the review in the attached supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-460-AC4
RC3: 'Comment on essd-2021-460', Anonymous Referee #3, 05 Sep 2022

As natural tracers of the water cycle, long term water stable isotope data are useful for hydrological and climatic studies. Chen et al. compiled observational precipitation oxygen isotope data from mainland China and eight iGCM simulation datasets from SWING2 to create a 148-year long monthly resolution precipitation oxygen isoscape for the mainland China. Rooted in the choice of the raw data, however, the usefulness of the generated data to the community is very limited. The processes and uncertainties regarding downscaling the model data from a resolution of 2 to 3 degrees to 50-60 km are also missing from the current manuscript. Therefore, the reliability and usefulness of the generated are highly doubted.

Major comments:

The selection of raw data:

Model outputs in the SWING2 project are from model versions of more than a decade ago, during which model physics and parameterizations have been updated for several versions. Such outdated data may only be used in the case of qualitative comparisons. While in such a quantitative data description paper, any new data generated from these raw data would inherent the quality issues in earlier model versions. Some of the model data from newer versions have already been made publicly available.

I find that the authors did not use simulation data from all models participated in the SWING2 project. However, the reason why the authors partly selected the datasets is not given.

Although the study is dedicated to the region of mainland China, adding station observations from neighboring regions would improve the quality and confidence of the generated data, especially at the margin of the mainland China.

During most of the time, data from only one model was used and there were limited or even no observations. For example, the authors used HadAM3 outputs for 1870-1957 when no observations are available. Several data quality issues arise for this period. During periods when there are observations, HadAM3 was not perform well over the study region. For the period of 1870-1957, the optimal method (among the methods evaluated by the authors) was also not used.

Methods:

My largest concern regarding the methods is the downscaling of the model outputs. I totally missed how the model outputs with resolutions of 2 to 3 degrees are downscaled to 0.5 degrees. The mainland China has a complex climatic and geographical characteristic, which could make the oxygen isotope ratios also have complex spatial variability at sub-pixel levels.

The consistency and the continuity of the data is also not guaranteed. The authors used CNN fusion for 1969-2007, and BCMs for the rest of 1870-2017. However, the different methods have different performance.

It is unclear how the observations are compared with iGCM outputs or the generated data: observations are at point level, iGCM simulations and the generated data are at grid level, comparisons are made at regional level.

Observations only cover a small part of the study period, and the authors may need to try to incorporate some proxy data.

Citation: https://doi.org/10.5194/essd-2021-460-RC3

Status: closed

RC1:
'Comment on essd-2021-460', Liheng Wang, 12 Apr 2022
The stable isotope composition of water is a very useful tracer to elucidate its formation history and cycle. The precipitation is an essential input for the hydrological cycle but it is very difficult to obtain continuous temporal and spatial variations of precipitation isotopes. This paper provides a high-resolution precipitation oxygen distribution in China based on IAEA long-term data and other datasets. These results will be of significant interest to the scientific community, particularly in hydrology. Overall, this review paper is scientifically robust given the available datasets. It is also reasonably written and well organized. I think the paper suits the reader of this journal and can be accepted for publication. I have some major comments and several minor comments for the authors to consider.

Major:

The topic of this paper is very interesting, in my opinion, it upgrades the data from a sparse point scale to a continuous regional area scale. I suggest that authors highlight this in the introduction section, and be more explicit about the meaning of the paper, which will help arouse the reader's interest and facilitate its further spread.

As we know, isotopic composition in precipitation also varies dramatically over time. Therefore, I think the temporal resolution is also important. What do you think about this? I think authors should state or discuss clearly the time resolution.

Minor:

Line 25: investigated —> shown?

Line 40 add reference: IAEA/WMO (2022). Global Network of Isotopes in Precipitation. The GNIP Database. Accessible at: https://nucleus.iaea.org/wiser

Line 63 watershed —> catchment

Figure 1 the unit (m) should be added to the legend.

IAEA also provided long-term data in Haikou, why isn’t it being used in your paper?

Line 96 There are many types of averages, are you using a monthly precipitation amount weighted average here?

Line 102 What is ï¼ would you like to give more information?

Line 219: Figure 3. I think the NW result is bad too. It should be discussed in detail.

Figure 5 – 8 What are SPR, SUM, AUT, and WIN? Jan- Mar is SPR? Authors should clearly indicate the months covered by each season. Table 2 also should be clearly mentioned.

Line 315-319. Author should add some references to prove your point.

Line338-40: You discussed TP again, why not move to Line302?

Line346: It's better to start a new paragraph so that it can be read more clearly.

Line 352-354: It's hard to see a significant trend.

Figure 9 Need to add y-axis labels.

Figure 10 I think it's very unclear to discuss why it is divided into these stages according to Figure 10. Based on methods? Or data? Authors should state it clearly.
Citation: https://doi.org/10.5194/essd-2021-460-RC1
- AC1: 'Reply on RC1', Jiacheng Chen, 18 Apr 2022
  
  We thank the reviewer for the valuable comments. We have responded to each point of the review in the attached supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-460-AC1
CC1:
'Comment on essd-2021-460', Nancy Swift, 26 Apr 2022
The authors create a high-resolution precipitation oxygen isoscape dataset for China by fusing eight iGCMs simulations and in-situ observations based on data fusion and bias correction techniques. I appreciate the authors’ great efforts to develop such a dataset, but I am quite dubious about the general content, and have large concerns about the novelty and quality of the dataset.
My main concerns are:
Data-quality and novelty:

It feels like a direct comparison of five commonly used fusion methods for developing a high-resolution dataset in China, without any new advanced fusion methods. Furthermore, the quality of the developed dataset is still questionable and unreliable due to its poor and insufficient present form. It seems to me that the intended novelty might be a high-resolution dataset.
The methods used to develop the dataset are unclear and not robust. For example, the sensitivity of model parameters are not evaluated and discussed; thus, the results are not robust. Why does CNN perform much better? Why is it set to a three-layer structure model? The observed data covers a short period and is not sufficient to train the model.
Using the interpretations of spatial pattern at the seasonally averaged scale (1969-2007) and the temporal pattern at the regional scale to validate the effectiveness and reliability of the data is not persuasive. Why not give us a comprehensive assessment at finer scale, such as time-series comparisons between the gridded simulations (50km) and in-situ observations at each station, and spatial patterns for each month. Without these comprehensive evaluations at finer scale, I am quite dubious about the data-quality and usefulness of this data set.
The presentation of dataset.

Introduction: The section of Introduction is not well written, and lacks to interconnect of the data it shares to and to show how it is valuable in relation to the Earth’s system. For example, readers should have a clear understanding of the motivation of this study, the purpose of creating such a dataset, which can be seen from a literature review of precipitation oxygen isoscape in hydrological and biogeochemical cycles. Then, followed by a detailed description of the available datasets, we could not find any description of the previously evaluated performance of iGCMs simulation in the Introduction. A description of the data-fusion method should also be added.
Data set and study area: Please add more details about the in-situ data (e.g., time-series of available in-situ data, number of data points at each station) in the supplementary material to justify the machine learning. For better visualization, a mesh of the iCGMs could be added in Fig.1
Methodology: This section is not well written, and should be clarified. For example, did you correct for iGCMs simulation bias at grid scale (grid by grid) or regional scale (using all gauges and all iGCMs within a specific region). How do these methods generate 50 km simulations from various iGCMs with different spatial resolutions?
Results and discussion: see above comments. Many published ESSD papers have demonstrated the uncertainties, limitations, advantages and 1-3 specific applications (for validation, evaluation, and analysis) of their dataset, which are also highly recommended for your paper.

For above reasons, I do not support its publication in the ESSD, without advanced approach or comprehensive evaluations of dataset at finer scale. Anyway, I still look forward to seeing the reviewers’ comments and editor’s decision.
Citation: https://doi.org/10.5194/essd-2021-460-CC1
- AC2: 'Reply on CC1', Jiacheng Chen, 31 May 2022
  
  We thank the reviewer for the valuable comments. We have responded to each point of the review in the attached supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-460-AC2
CC2:
'Comment on essd-2021-460', Donal Logue, 28 Apr 2022

General comments:
Obtaining a high-resolution long-term precipitation oxygen isoscape dataset can be critical for relevant hydrological studies. This study presented a first attempt to solve this issue, really appreciate, but from my point of view, it is still a rather premature dataset and have limited value. The authors downscaled and fused eight iGCMs precipitation oxygen isoscape using five different methods from a coarse spatial resolution (~2/3 degree) to a higher spatial resolution (0.5 degree). However, the work is not innovative and no important and robust findings were obtained. The results do not convince me since this method highly depends on the training data, which cover a short period, are unevenly distributed across regions, and insufficient to train model. Moreover, as we know, the precipitation oxygen isoscape is highly dependent on local climate conditions, terrain factors, as well as large-scale atmospheric and local circulation. No such physical-based ancillary data were used in this study, which limited the further applications of produced 0.5-degree data. The sub-pixel spatial patterns within a coarse pixel changes, but the current methodology cannot get this information. For the implementation of models, uncertainty of the datasets resulting from the model structures, parameters, training and testing strategies is not even discussed. Data-quality assessment at finer scale is poor presented. I have concerns about the reliability of spatial-temporal variations in your new data product at fine resolution.
In conclusion, this is a good attempt to generate a high-resolution dataset based on the fusion of in-situ data and iGCMs simulations, however, due to the above-mentioned issues, I don’t think this dataset meet the high standard of ESSD.

Citation: https://doi.org/10.5194/essd-2021-460-CC2
- AC3: 'Reply on CC2', Jiacheng Chen, 31 May 2022
  
  We thank the reviewer for the valuable comments. We have responded to each point of the review in the attached supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-460-AC3
RC2:
'Comment on essd-2021-460', Anonymous Referee #2, 14 May 2022
As one of the key tracers of hydroclimate change, precipitation isotopes have very important research significance. According to the current situation of less observational data, the authors used iGCMs to obtain high-resolution precipitation isotope data over the past 148 years, which will provide important data support for the study of precipitation isotopes and hydroclimate change. I think this research is meaningful, but there are still some problems in the article, I think it needs to be further elaborated before accepting publication.

There is a problem with regional division. For example, Yunnan Province is a typical southwestern region of China, and its climate is dominantly influenced by the Indian summer monsoon, which is different from the region of southeastern China where is mainly influenced the East Asian summer monsoon. How can it be divided into southeastern regions? Generally, the Indian summer monsoon precipitation oxygen isotope values during JJAS are lower than the East Asian summer monsoon rainfall. I don’t think current regional division is scientific.

There are some GNIP stations with long-term precipitation isotope monitoring data, which can be used to compare long-term changes with the simulation results, such as the Hong Kong station. I suggest that the authors can compare the Hong Kong station and other stations with 8-10 years of monitoring data with the simulated isotope records. This comparison can be used to verify the reliability of the simulation.

At present, the altitudes of each monitoring site vary greatly. When the author uses the monitoring data of each site for spatial analysis and compares them with the simulation results, did the author consider the effect of altitude on precipitation isotopes to calibrate? For example, according to the relationship between altitude and precipitation isotopes at each site or the large region scale, first calibrate the precipitation isotope data of all stations to the same altitude, and then compare it with the simulated data. Because the simulation precipitation isotope results are at the same altitude, if no calibration is performed, the spatial comparison of the simulated and monitored precipitation isotope data will inevitably not be the real results. In the current manuscript, it seems that the author has not calibrated, and it is suggested that the author add relevant correction processes or solutions.
Citation: https://doi.org/10.5194/essd-2021-460-RC2
- AC4: 'Reply on RC2', Jiacheng Chen, 31 May 2022
  
  We thank the reviewer for the valuable comments. We have responded to each point of the review in the attached supplement.
  
  Citation: https://doi.org/10.5194/essd-2021-460-AC4
RC3: 'Comment on essd-2021-460', Anonymous Referee #3, 05 Sep 2022

As natural tracers of the water cycle, long term water stable isotope data are useful for hydrological and climatic studies. Chen et al. compiled observational precipitation oxygen isotope data from mainland China and eight iGCM simulation datasets from SWING2 to create a 148-year long monthly resolution precipitation oxygen isoscape for the mainland China. Rooted in the choice of the raw data, however, the usefulness of the generated data to the community is very limited. The processes and uncertainties regarding downscaling the model data from a resolution of 2 to 3 degrees to 50-60 km are also missing from the current manuscript. Therefore, the reliability and usefulness of the generated are highly doubted.

Major comments:

The selection of raw data:

Model outputs in the SWING2 project are from model versions of more than a decade ago, during which model physics and parameterizations have been updated for several versions. Such outdated data may only be used in the case of qualitative comparisons. While in such a quantitative data description paper, any new data generated from these raw data would inherent the quality issues in earlier model versions. Some of the model data from newer versions have already been made publicly available.

I find that the authors did not use simulation data from all models participated in the SWING2 project. However, the reason why the authors partly selected the datasets is not given.

Although the study is dedicated to the region of mainland China, adding station observations from neighboring regions would improve the quality and confidence of the generated data, especially at the margin of the mainland China.

During most of the time, data from only one model was used and there were limited or even no observations. For example, the authors used HadAM3 outputs for 1870-1957 when no observations are available. Several data quality issues arise for this period. During periods when there are observations, HadAM3 was not perform well over the study region. For the period of 1870-1957, the optimal method (among the methods evaluated by the authors) was also not used.

Methods:

My largest concern regarding the methods is the downscaling of the model outputs. I totally missed how the model outputs with resolutions of 2 to 3 degrees are downscaled to 0.5 degrees. The mainland China has a complex climatic and geographical characteristic, which could make the oxygen isotope ratios also have complex spatial variability at sub-pixel levels.

The consistency and the continuity of the data is also not guaranteed. The authors used CNN fusion for 1969-2007, and BCMs for the rest of 1870-2017. However, the different methods have different performance.

It is unclear how the observations are compared with iGCM outputs or the generated data: observations are at point level, iGCM simulations and the generated data are at grid level, comparisons are made at regional level.

Observations only cover a small part of the study period, and the authors may need to try to incorporate some proxy data.

Citation: https://doi.org/10.5194/essd-2021-460-RC3

Jiacheng Chen et al.

Data sets

Precipitation oxygen isoscape for mainland China from 1870 to 2017 generated based on data fusion and bias correction of iGCMs simulations Chen, Jiacheng; Chen, Jie; Zhang, Xunchang J.; Peng, Peiyi; Risi, Camille https://doi.org/10.5281/zenodo.5703811

Jiacheng Chen et al.

Viewed

Total article views: 876 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
698	146	32	876	6	9

HTML: 698
PDF: 146
XML: 32
Total: 876
BibTeX: 6
EndNote: 9

Views and downloads (calculated since 01 Apr 2022)

Month	HTML	PDF	XML	Total
Apr 2022	240	48	15	303
May 2022	148	26	9	183
Jun 2022	71	17	2	90
Jul 2022	51	9	0	60
Aug 2022	53	16	2	71
Sep 2022	48	13	2	63
Oct 2022	35	6	1	42
Nov 2022	30	5	0	35
Dec 2022	14	5	1	20
Jan 2023	8	1	0	9

Cumulative views and downloads (calculated since 01 Apr 2022)

Month	HTML	PDF	XML	Total
Apr 2022	240	48	15	303
May 2022	148	26	9	183
Jun 2022	71	17	2	90
Jul 2022	51	9	0	60
Aug 2022	53	16	2	71
Sep 2022	48	13	2	63
Oct 2022	35	6	1	42
Nov 2022	30	5	0	35
Dec 2022	14	5	1	20
Jan 2023	8	1	0	9

Viewed (geographical distribution)

Total article views: 818 (including HTML, PDF, and XML) Thereof 818 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 28 Jan 2023

Short summary

To make full use of the advantages of isotope observations and simulations, this study generates a new dataset by integrating multi-GCM data based on data fusion and bias correction methods. This dataset contains monthly δ¹⁸O_p over mainland China for the 1870–2017 period with a spatial resolution of 50–60 km. The built isoscape shows similar spatial and temporal distribution characteristics to observations, which is reliable and useful to extend the time and space of observations in China.


Total:	0
HTML:	0
PDF:	0
XML:	0