High-resolution datasets for lake level changes in the Qinghai-Tibetan Plateau from 2002 to 2021 using multi-altimeter data

Chen, Jiaming; Liao, Jingjuan; Lou, Yanhan; Ma, Shanmu; Shen, Guozhuang; Zhang, Lianchong

doi:https://doi.org/10.5194/essd-2022-313

Preprints

https://doi.org/10.5194/essd-2022-313

Preprints

18 Oct 2022

| 18 Oct 2022

Status: a revised version of this preprint is currently under review for the journal ESSD.

High-resolution datasets for lake level changes in the Qinghai-Tibetan Plateau from 2002 to 2021 using multi-altimeter data

Jiaming Chen, Jingjuan Liao, Yanhan Lou, Shanmu Ma, Guozhuang Shen, and Lianchong Zhang

Abstract. The Qinghai Tibet Plateau (QTP), known as the Roof of the World and the Water Tower of Asia, has the largest number of lakes in the world, and because of its high altitude and near absence of disturbances by human activity, the plateau has long been an important site for studying global climate change. Hydrological stations cannot be readily set up in this region, and in situ gauge data are not always publicly accessible. Satellite radar altimetry has become a very important alternative to in situ observations as a source of data. Estimation of the water levels of lakes via radar altimetry is often limited by temporal and spatial coverage, and, therefore, multi-altimeter data are often used to monitor lake levels. Restricted by the accuracy of waveform processing and the interval period between different altimetry missions, the accuracy and the sampling frequency of the water level series are typically low. By processing and merging data from eight different altimetry missions, the developed datasets provided the water level changes for 362 lakes (larger than 10 km²) in the QTP from 2002 to 2021. The period for the lake level change series, which affords high accuracy, can be much longer for many lake systems. The present datasets and associated approaches are valuable for calculating the changes in lake storage, trend analyses of the lake levels, short-term monitoring of the overflow of lakes, flooding disasters on the plateau, and the relationships between changes in the lake ecosystems and changes in the water resources.

Received: 06 Sep 2022 – Discussion started: 18 Oct 2022

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Jiaming Chen, Jingjuan Liao, Yanhan Lou, Shanmu Ma, Guozhuang Shen, and Lianchong Zhang

Status: final response (author comments only)

RC1: 'Comment on essd-2022-313', Anonymous Referee #1, 17 Nov 2022

This study provides the time series of water level for lakes in the Qinghai-Tibetan Plateau between 2002 and 2021 using altimeter data from Envisat, ICESat-1, CryoSat-2, Jason-1, Jason-2, Jason-3, SARAL, and Sentinel-3A. The water level data in 2002-2021 provided by this study is not new, and have been reported by couple of previous studies. This study did not present well such as time series of lake level and not story focused such as the mention the discharge without close relation with study study. Moreover, the authors did not know the background information of lakes over the Tibetan Plateau well. I can not recommend the publication of this manuscript (dataset).

1) Although this study provides the time series of water level of lakes in the Qinghai-Tibetan Plateau from eight altimetry products, this dataset is not new compared with published studies, especially in a limited period (2002-2021). Hydroweb and other websites have provided open access lake level data since 1992, which has covered the altimetry data used in this study.

2) The presentation of this study is not good such as Figures 3 and 4. Flowchart 1 and 5 should be combined together. For a scientific paper, the figures should be drawn by a scientific standard. Moreover, the offset among the different altimetry data was addressed? How?

3) What is the difference of boundaries between Qinghai-Tibetan Plateau and Tibetan Plateau? How the comparison of time series of altimetry data and in-situ? Why the streamflow and discharge data are used, but the analysis of water level and advantage of your study are not clear?

Citation: https://doi.org/10.5194/essd-2022-313-RC1

AC1: 'Reply on RC1', Jingjuan Liao, 27 Nov 2022

The authors would like to express thanks to the anonymous reviewer for his voluntary work and the constructive comments to improve this manuscript. All of the comments are of great benefit to us. During the past few days, we did much work to revise the manuscript according to the reviewer’s comments. All of the comments have been addressed as following:

RC1: 'Comment on essd-2022-313', Anonymous Referee #1, 17 Nov 2022

Q: This study provides the time series of water level for lakes in the Qinghai-Tibetan Plateau between 2002 and 2021 using altimeter data from Envisat, ICESat-1, CryoSat-2, Jason-1, Jason-2, Jason-3, SARAL, and Sentinel-3A. The water level data in 2002-2021 provided by this study is not new, and have been reported by couple of previous studies. This study did not present well such as time series of lake level and not story focused such as the mention the discharge without close relation with study study. Moreover, the authors did not know the background information of lakes over the Tibetan Plateau well. I can not recommend the publication of this manuscript (dataset).

Reply: Although the water level data provided by this study is not new, we provide the largest number of monitored lakes among the available studies. In the revised version, we added the analysis of the variation of lake level in different basin, while the discharge mentioned in the manuscript is mainly to illustrate the application of lake level data, which refers to the fact that we can explore the regulation of the rivers associated with the lake using lake level data.

Q: 1) Although this study provides the time series of water level of lakes in the Qinghai-Tibetan Plateau from eight altimetry products, this dataset is not new compared with published studies, especially in a limited period (2002-2021). Hydroweb and other websites have provided open access lake level data since 1992, which has covered the altimetry data used in this study.

Reply: These eight altimetry products are already used in published studies, but the number of lakes monitored based on them is very limited, as can be seen in the following table.

Paper	No. of Lakes	Period	Data Source	Dataset Public or not
Jiang et al. (2017)	70	2003-2015	IceSat-1, Cryosat-2	N
Zhang et al. (2017)	68	1989-2015	IceSat-1, Landsat	N
Li et al. (2017)	167	2002-2012	IceSat-1, Envisat	N
Hwang et al. (2019)	59	2003-2016	Jason-2/3, SARAL, IceSat-1, Cryosat-2	N
Li et al. (2019)	52	2000-2017	Jason-1/2/3, Envisat, Cryosat-2, IceSat-1	Y
Zhang et al. (2019)	62	2003-2018	IceSat-1/2	Y
Hydroweb (Cretaux et al. 2011)	36	1993-2022	ERS-2, Envisat, T/P, IceSat-1, SARAL, Jason-1/2/3, Cryosat-2, Sentinel-3A	Y
DAHITI (Schwatke et al. 2015)	62	2003-2022	ERS-2, Envisat, SARAL, Sentinel-3A, Cryosat-2, IceSat-1, Jason-2/3,	Y
This Study	362	2002-2021	Envisat, SARAL, IceSat-1, Cryosat-2, Jason-1/2/3, Sentinel-3A	Y

As we know, a large number of lakes exist on the Tibetan Plateau, and monitoring the water levels of these lakes is very important for understanding the water cycle on the this plateau. Our study monitored the largest number of lakes with an area greater than 10 km² compared to previous studies. We believe that considering more lakes will be useful to find more details, driving mechanisms, and patterns of changes in the Tibetan Plateau. So this is also the objective of this manuscript.

Q: 2) The presentation of this study is not good such as Figures 3 and 4. Flowchart 1 and 5 should be combined together. For a scientific paper, the figures should be drawn by a scientific standard. Moreover, the offset among the different altimetry data was addressed? How?

Reply: Thank for your comments. We have combined the Figure 3 and 4, and drawn the figure again, please see the revised manuscript. Since Flowchart 1 and 5 represent different processes for altimetry data, Flowchart 1 is mainly the waveform retracking processing of altimetry data, while process 5 is mainly the fusion processing of water level data extracted from multi-source altimeters, which are different and not suitable for merging, so the two flowcharts are retained.

In this manuscript, the main merged method removes the offset between different altimetry data by subtracting the mean discrepancy obtained during the overlap period.

Thus, we will pick the dynamic reference time series to make the mergerd time series as long as possible for each time in case there are no overlap period.

On the other hand, it also exists some lakes with no overlap period when merging ICEsat-1 and Cryosat-2. In this case, we will consider using a combined linear-periodic-residual model (Liao et al., 2014) to simulate and forecast lake-level time series in the no-overlap period, and then make it possible to obtain the offset between ICEsat-1 and Cryosat-2. These details are presented in section 3.2.

Q: 3) What is the difference of boundaries between Qinghai-Tibetan Plateau and Tibetan Plateau? How the comparison of time series of altimetry data and in-situ? Why the streamflow and discharge data are used, but the analysis of water level and advantage of your study are not clear?

Reply: There is no difference of boundaries between Qinghai-Tibetan Plateau and Tibetan Plateau. In China, people used to call Qinghai-Tibetan Plateau (QTP), but internationally, it is used to call Tibetan Plateau (TP). To avoid misunderstanding, we have revised as the Tibetan Plateau in this manuscript.

Due to the unknown datum of in situ data, here we consider comparing the water level anomaly between in situ data and lake level in this study by removing the mean value over the validation period. In the revised vision, in order to show the validate results of lake level in this study, we added a figure comparing the in situ data with the lake level in this study.

The streamflow and discharge data are used to show that lake levels can be used to explore the regulation of the rivers associated with the lake, and this is just one case of the application of lake levels. In addition, we added the analysis of water level changes in lakes in different basin, so it further indicates that our study is much clear in reflecting the spatial and temporal variability of lakes on the Tibetan Plateau.

Citation: https://doi.org/10.5194/essd-2022-313-AC1

CC1:
'Comment on essd-2022-313', Nancy Swift, 08 Dec 2022

I totally agree with the first Anonymous Referee. Below I would like to introduce some interesting works:
1) Liu, J., Fang, P., Que, Y., Zhu, L.-J., Duan, Z., Tang, G., Liu, P., Ji, M., and Liu, Y.: A dataset of lake-catchment characteristics for the Tibetan Plateau, Earth Syst. Sci. Data, 14, 3791–3805, https://doi.org/10.5194/essd-14-3791-2022, 2022.
This study also provides water level time series of lakes from Li et al. (2019), Zhang et al. (2021), and Xu et al. (2022).
2) Tourian, M. J., Elmi, O., Shafaghi, Y., Behnia, S., Saemian, P., Schlesinger, R., and Sneeuw, N.: HydroSat: geometric quantities of the global water cycle from geodetic satellites, Earth Syst. Sci. Data, 14, 2463–2486, https://doi.org/10.5194/essd-14-2463-2022, 2022.

This study provides water level, area and storage variation time series of global lakes/reservoirs/rivers from multisource datasets.

3) Hou, J., Van Dijk, A. I. J. M., Renzullo, L. J., and Larraondo, P. R.: GloLakes: a database of global lake water storage dynamics from 1984 to present derived using laser and radar altimetry and optical remote sensing, Earth Syst. Sci. Data Discuss. [preprint], https://doi.org/10.5194/essd-2022-266, in review, 2022.

This study provides global lake water storage dynamics using water area and water level from multiple altimeters. However, there are critical comments from referees.

Regarding the increase in the spatial coverage of the lakes you claimed, I doubt its usefulness as there is very little data available for these small lakes. This is the main reason Hydroweb/other databases (also using multisource satellite altimeters) do not provide time series of water levels for these lakes. This issue may be addressed from future SWOT mission.

Overall, your datasets are neither new nor exhaustive and complete (such as Tourian et al. 2022 and Liu et al. 2022)! I could not see any novelty in your work (new method such as Li et al. 2019 addressed the inter-satellite biases) and a comprehensive validation. For above reasons, I do not support its publication in ESSD.

Citation: https://doi.org/10.5194/essd-2022-313-CC1
- AC2:
  'Reply on CC1', Jingjuan Liao, 12 Jan 2023
  
  Thanks for the comments. Actually, the number of lakes monitored based on them is very limited, as can be seen in the table 1 in revised version. Our study monitored the largest number of lakes with an area greater than 10 km² in Tibetan Plateau (not Global) compared to previous studies. We believe that considering more lakes will be useful to find more details, driving mechanisms, and patterns of changes in the Tibetan Plateau. This is also the main objective of this manuscript.
  It is not necessary to say some one lake level time series is usefulness, even though some lakes are only available on Cryosat-2, because it could also give us a clear linear trend at recently 10 years. Once we monitored more lakes as much as possible, we could find some interesting and different spatial driving mechanisms, and patterns of changes in the Tibetan Plateau. Such as most lakes located in the southeastern Qiangtang Basin and near the Nyainqen Tanglha Mountain shows a decreasing trend which is different from patterns of changes of the whole Qiangtang basin. If we don’t try to monitor these lakes, we cannot find this pattern of changes. Moreover, in the southern Tibetan Plateau (in Brahmaputra Basin), we can find 8 lakes shown a decreasing trend, it is no more than 4 lakes were monitored in this area in previous area. Comparing with this, it is obvious that the dataset from this manuscript can be more suitable to explore the spatial driving mechanisms, and patterns of changes in the Tibetan Plateau. You can see details of the spatio-temporal analysis of changes in lake levels in the TP in revised version.
  Additionally, we use our in-home developed retracker to retrack all the altimeters. And we also develop a method (a combined linear-periodic-residual model) to merge the lake level time series for those lakes without overlap period between two different altimeters. We don’t mention too much details in this manuscript, because we have published related methodology papers (Chen et al., 2020; Liao et al., 2014), and this is a dataset description manuscript.
  
  Citation: https://doi.org/10.5194/essd-2022-313-AC2
  - CC2: 'Reply on AC2', Nancy Swift, 14 Jan 2023
    
    The authors contend that a high-resolution dataset might be an intended novelty. Of the 362 lakes I studied, 190 only had time-series from a single altimeter (CryoSat-2 and/or ICESat-2). Lake measurements are only available once per year on average, limiting their use to a wide range of applications. Furthermore, I noticed you provided operational Sentinel-3A and Jason-3 data as of 2019.08. Shouldn't this data be updated to 2021 (based on your title) or even 2023? Moreover, why not include other satellite missions like ICESat-2 and Sentinel-3B? Only eight lakes with in situ data were assessed for data-quality. You have not provided a detailed evaluation of your data product, so I am seriously concerned about its reliability.
    Despite mentioning this reference in my previous comment, the authors failed to include it, HydroSat (Tourian et al., 2022 ESSD). Hydrosat provides water level time series for global rivers, lakes, and reservoirs using Envisat, Saral, ICESat-2, CryoSat-2, Jason-1/-2/-3, and Sentinel-3A. Tourian presented a comprehensive overview of all methods for dealing with inter-satellite bias (Section 2.2.3). I do not see any novelty in your method for addressing inter-satellite biases. A new methodology has been proposed by Li et al. (2019), which allows them to publish their high-resolution datasets in ESSD. Despite three years passing, I am unable to discover any novelty in your work except for a high rate of similar follow-up. The application of the developed dataset demonstrated in the manuscript is not the novelty of this work and the overall framework is highly similar to Li et al. (2019).
    In terms of its general structure, the framework is very similar to Li et al. (2019) and does not provide any significant improvements over the existing works, in line with ESSD standards. As such, I would suggest a rejection.
    References:
    Tourian, M. J., Elmi, O., Shafaghi, Y., Behnia, S., Saemian, P., Schlesinger, R., and Sneeuw, N.: HydroSat: geometric quantities of the global water cycle from geodetic satellites, Earth Syst. Sci. Data, 14, 2463–2486, https://doi.org/10.5194/essd-14-2463-2022, 2022.
    Li, X., Long, D., Huang, Q., Han, P., Zhao, F., and Wada, Y.: High-temporal-resolution water level and storage change data sets for lakes on the Tibetan Plateau during 2000–2017 using multiple altimetric missions and Landsat-derived lake shoreline positions, Earth Syst. Sci. Data, 11, 1603–1627, https://doi.org/10.5194/essd-11-1603-2019, 2019.
    
    Citation: https://doi.org/10.5194/essd-2022-313-CC2

RC2: 'Comment on essd-2022-313', Anonymous Referee #2, 26 Dec 2022

Line 13: Here is for global climate change. But the text cannot discuss the relation between lake level change and global climate change.
Line 16: What about the detail altimetry satellite missions?
Line 19: Here the time spans 2002 to 2021. Are all lakes’ levels in this time span?
Section 1: There are many literatures about the lake level changes in QTP, which should be further summarized and generalized.
Line 72: Here there are 364 lakes. But there are 262 lakes in line 19.
Table 1: One altimetry satellite can only pass some lakes. Some lakes can be covered by two or three altimetry satellite mission. How to process these conditions to precisely determine one level series for one lake? How to get the lake level series of all lakes from 2002 to 2021?
Table 2: What unit is used for coordinate? 1985 should be the National Height Datum of China.
(1): How to get these corrections? How to improve these corrections?
Section 3.1.1: How to verify the effectiveness and accuracy of the retracking method?
Section 3.1.2: Here is noise footprints. How about the abnormal footprint?
Section 3.1.3: How to process the gross errors in time series?
Section 3.2: Coordinate system transfer and coordinate frame transfer all should be made.
(5): How to determine p in the Eq. ?
Section 4: How about resolution?
Table 3: RMS of Zhari Namco is 0.25m. But RMS is about 10.1cm in Sun et al. (Detecting lake level change from 1992 to 2019 of Zhari Namco in Tibet using altimetry data of TOPEX/Poseidon and Jason-1/2/3 missions. Frontiers in Earth Science, 2021, 9:640553, https://doi.org/10.3389/feart.2021.640553). Wang et al. (Robust, long-term lake level change from multiple satellite altimeters in Tibet: observing the rapid rise of Ngangzi Co over a new wetland. Remote Sensing, 11: 558, doi: 10.3390/rs11050558.) also shown the more precise lake levels.
Section 5: How about the physical mechanism? Why not the applications to climate change?

Citation: https://doi.org/10.5194/essd-2022-313-RC2

AC3: 'Reply on RC2', Jingjuan Liao, 12 Jan 2023

Line 13: Here is for global climate change. But the text cannot discuss the relation between lake level change and global climate change.

Reply: Thanks for the suggestions, but here we are not discussing the relation between lake level change and global climate change. Just an introduction for the Tibetan Plateau and describe the importance of lakes in Tibetan Plateau.

Line 16: What about the detail altimetry satellite missions?

Reply: Thanks for the suggestions, the altimetry missions we used in this manuscript is added in the revised version.

Line 19: Here the time spans 2002 to 2021. Are all lakes’ levels in this time span?

Reply: Thanks for the suggestions, not all the lakes staring from 2002, among all of this, 168 lakes water level staring from 2010. The details also added into revised version.

Section 1: There are many literatures about the lake level changes in QTP, which should be further summarized and generalized.

Reply: Thanks for the suggestions, we added related literatures in past 5 years and summarized in the table 1.

Table 1 Comparison of this study with previous studies

Reference	No. of Lakes	Period	Data Source	Dataset Public or not
Jiang et al. (2017)	70	2003-2015	IceSat-1, Cryosat-2	N
Zhang et al. (2017)	68	1989-2015	IceSat-1, Landsat	N
Li et al. (2017)	167	2002-2012	IceSat-1, Envisat	N
Hwang et al. (2019)	59	2003-2016	Jason-2/3, SARAL, IceSat-1, Cryosat-2	N
Li et al. (2019)	52	2000-2017	Jason-1/2/3, Envisat, Cryosat-2, IceSat-1	Y
Zhang et al. (2019)	62	2003-2018	IceSat-1/2	Y
Hydroweb	36	1993-2022	ERS-2, Envisat, T/P, IceSat-1, SARAL, Jason-1/2/3, Cryosat-2, Sentinel-3A	Y
DAHITI	62	2003-2022	ERS-2, Envisat, SARAL, Sentinel-3A, Cryosat-2, IceSat-1, Jason-2/3,	Y
This Study	362	2002-2021	Envisat, SARAL, IceSat-1, Cryosat-2, Jason-1/2/3, Sentinel-3A	Y

Line 72: Here there are 364 lakes. But there are 262 lakes in line 19.

Reply: Thanks for the suggestions, we have revised it, the correct number is 362. We also scrutiny the whole text, and revised this mistake.

Table 1: One altimetry satellite can only pass some lakes. Some lakes can be covered by two or three altimetry satellite mission. How to process these conditions to precisely determine one level series for one lake? How to get the lake level series of all lakes from 2002 to 2021?

Reply: Thanks for the comments.

After constructing the time series for each altimetry over related lakes. We can follow the section 3.2 to fuse the multi-altimeter time series.

Specially, we first merged the two products with the longest period for the time series and chose the altimeter-derived water level with the longer time series as the baseline. Then systematic biases between another altimeter and the baseline will be removed by subtracting the mean discrepancy during the overlap period compared with the reference series (Lee et at., 2011; Kropáček et al., 2012) according to Eq. (4). Then, the same process was applied to the remaining products and the merged products connecting the three altimeters.

Additionally, some lakes cannot merged successfully (if just pass ICEsat-1 and Cryosat-2 before 2013), we will try to use a combined linear-periodic-residual model (Liao et al., 2014) to simulate and forecast the lake-level time series in the no-overlap period to merge the two altimeters with no overlap period.

We added some details to make it easier to follow in revised version.

Unfortunately, this manuscript only estimate lake level has valid observations from multi-altimetry, so not all the 364 lakes for time series from 2002 to 2021, 168 lakes for the time series from 2010 to 2021.

Table 2: What unit is used for coordinate? 1985 should be the National Height Datum of China.

Reply: Thanks for the suggestions, the unit for coordinate is degree. We revised the table notes for coordinates and 1985.

(1): How to get these corrections? How to improve these corrections?

Reply: Thanks for the comments. With the exception for retracking correction, all the corrections are included in the altimetry data product. We just need to improve the retracking corrections which is the most important to decide the accuracy of lake level. More detail can be seen in section 3.1.1.

Section 3.1.1: How to verify the effectiveness and accuracy of the retracking method?

Reply: Thanks for the suggestions. Actually, the retracking method we used here is the automatic multiscale-based peak detection retracker (AMPDR), having a special paper for it (Chen et al., 2020). We try to compare several lake level time series from altimetry with in-situ time series and existing product time series. The effectiveness and accuracy of this retracker has already discussed in detail in the paper above.

Section 3.1.2: Here is noise footprints. How about the abnormal footprint?

Reply: Thanks for the suggestions. The abnormal footprints are already detected in section 3.1.1, we call as unavailable off-nadir observation, which will be removed during retracking.

The lake level from invalid off-nadir observation usually shows a large bias with the DistanceThres. So, if the difference is larger than the half of the range window, it may not provide the signals from water surface and this observation will be regarded as an invalid off-nadir point (using Sentinel-3 as an example, it is 1/2∗128∗0.4684). (Chen et al., 2020)

Section 3.1.3: How to process the gross errors in time series?

Reply: Thanks for the suggestions. Here we used a state-space model (Nielsen et al., 2015), the gross errors can be obtained from their models. The obtained time series are not easily averaging, they considering continuous time steps and a hypothetical error model. Then according to the Laplace estimation, the mean value and error will be calculated from the model. More detail can be seen in Nielsen et al. (2015).

Section 3.2: Coordinate system transfer and coordinate frame transfer all should be made.

Reply: Thanks for the suggestions. Coordinate system transfer and coordinate frame transfer are all made in this manuscript. All the data will be transferred into WGS84/EGM 2008.

(5): How to determine p in the Eq. ?

Reply: Thanks for the suggestions. p stands for the number of periodic components. This can be determined by the reality. Such as considering 10 years periodic, 5 yeasr periodic, 2 years periodic, and 1 year periodic, then p equals to 4.

Section 4: How about resolution?

Reply: Thanks for the suggestions. Due to the complexity of fusion of multi-altimeter, it is hard to give a spatial resolution for the data set. The high resolution in title means high spatial coverage and large numbers of lakes.

Table 3: RMS of Zhari Namco is 0.25m. But RMS is about 10.1cm in Sun et al. (Detecting lake level change from 1992 to 2019 of Zhari Namco in Tibet using altimetry data of TOPEX/Poseidon and Jason-1/2/3 missions. Frontiers in Earth Science, 2021, 9:640553, https://doi.org/10.3389/feart.2021.640553). Wang et al. (Robust, long-term lake level change from multiple satellite altimeters in Tibet: observing the rapid rise of Ngangzi Co over a new wetland. Remote Sensing, 11: 558, doi: 10.3390/rs11050558.) also shown the more precise lake levels.

Reply: Thanks for the suggestions. For the RMSE of Zhari Namco, there is two possible reasons for this problem, one is that Zhari Namco is available on the eight altimeters in this manuscript, the data quality of Envisat is not stable, another is that fusion many different altimeter will also introduce noise. Actually, if we just consider Cryosat-2, Sentinel-3, Jason-2, and Jason-3, the RMSE is just 9.9 cm (as reported in Chen et al., 2020).

For the Ngangzi Co, no in situ gauge data are available in Ngangzi Co, so they are considering inter-compare with ICEsat and SARAL. In this manuscript, we just consider to compare with in-situ dada or existing product.

Section 5: How about the physical mechanism? Why not the applications to climate change?

Reply: There are many factors influencing lake level changes on the Tibetan Plateau, including Precipitation, and glacier and snow melt and degradation of permafrost brought by temperature changes, etc. This manuscript, as a dataset paper, mainly shows the changes of lake level and provides the basic data of lake level changes, so the physical mechanisms of lake level changes are not explored here, which will be further explored in future work.

Lake level changes on the TP are an indicator of climate change in the region, which has been addressed in many papers (such as Gao et al., 2013; Hwang et al., 2016; Hwang et al., 2019; Jiang et al., 2017), but here this manuscript does not expand on the application, but only points out the importance in the introduction.

Citation: https://doi.org/10.5194/essd-2022-313-AC3

RC3:
'Reply on AC3', Anonymous Referee #2, 14 Jan 2023

Only one altimetry satellite passes some lakes. But there are many lakes covered by two or three altimetry satellites. How to combine several satellite altimeters's data to optimize lake level changes? How to imporve lake level data quality further?

Citation: https://doi.org/10.5194/essd-2022-313-RC3
- AC4: 'Reply on RC3', Jingjuan Liao, 18 Jan 2023
  
  Q. Only one altimetry satellite passes some lakes. But there are many lakes covered by two or three altimetry satellites. How to combine several satellite altimeters's data to optimize lake level changes? How to imporve lake level data quality further?
  Reply: Thanks for the comments.
  For combining several satellite altimetry lake level time series, we need to remove the systematic biases first. The systematic biases between another altimeter and the baseline altimeter will be removed by subtracting the mean discrepancy during the overlap period compared with the reference series. For several altimeter data available on one lake in some day, we will try to consider the average value as the final water level.
  Additionally, some lakes cannot merge successfully (if just pass ICEsat-1 and Cryosat-2 before 2013), a combined linear-periodic-residual model (Liao et al., 2014) is used to simulate and forecast the lake-level time series in the no-overlap period to make it possible for merging the two altimeters with no overlap period.
  In order to improve lake level data quality further, we also intended to develop a new model to consider different altimeter data together in the future, regarding the systematic biases and error (assuming followed by some kind of errors model) as a fitting parameter. Using this way will be helpful for improve the data quality for multi-altimeter.
  
  Citation: https://doi.org/10.5194/essd-2022-313-AC4

RC4:
'Comment on essd-2022-313', Anonymous Referee #3, 28 Apr 2023

The manuscript from Chen et al. presents the development and validation of a water elevation time series database for 362 lakes in the Qinghai-Tibetan Plateau from satellite multi-mission altimeters.
General comment:
The manuscript is easy to read and the methodology to derive water elevation is adapted, even if more information is needed (see specific comments below). However, I have the following main concerns:
- The database is validated only for 8 lakes over 362. No information on the area of these lakes are provided, nor their locations, nor their hydrological regime. So, it’s not easy to know if they are representative of many lakes within the 362 lakes in the database.
- For the remaining 354 lakes, no regional consistency analysis is done between neighbor lakes. It could be a way of cross-validating the database. Similarly, for each lake, no consistency between time series from different missions is done. It would also be a way to detect if some time series might be erroneous or not compared to measurements from other missions. The intermission bias is a good way also to check if the different missions are observing the same target (if one mission provide water elevations multiples decameters above/below other missions, then they are not observing the same target).
- The comparison with other altimetry database is pretty weak. Lake trend is compared between the proposed database and the Hydroweb database. Why not comparing directly water elevations? Other altimetry databases (like DAHITI and G-REALM) should also be considered.
- Multiple satellite missions are used, which have different type of sensors. For example, ICESat is a lidar altimeter, which is quite different from nadir radar altimeters (both in vertical accuracy, sensitivity to clouds…). It is never explained, nor discussed. Pros and cons from lidar and nadir radar altimeters should be provided. For example, I would expect more accurate, but less measurements in time, from ICESat than from nadir radar altimeters.
- Besides, there are some errors from nadir radar altimeters that are not discussed, but could have a huge impact on the database. This type of altimeters, in closed-loop tracking mode or with erroneous onboard DEM value in open-loop tracking mode, could lock their tracking window on the top of the surrounding topography near the lakes. If this topography is quite high compared to the lake (>tracking window size), the waveform will not sample the lake surface elevation. So, no matter how efficient is the retracking algorithm, it is not possible to retrieve the lake surface elevation. This point should be discussed in the manuscript. In addition to this type of error, when the lake is frozen, nadir radar altimeter could provide erroneous data. How is it delat with in the database?
- There are more and more published papers using satellite lidar data (ICESat, ICESat-2 and GEDI), see for example Luo et al. (2021) who studied 221 lakes in the QTP. How does your database compare to these studies?
- Very few lake water elevation time series are presented (only three time series are shown, respectively in Figure 2, 3 and 4). They have rather a poor temporal sampling (figure 4) or strange elevation dynamic (figure 2).
- At the database repository, it is written that ‘196 lakes have the series from 2002 to 2021 in the 02-10 folder and 168 lakes have the series from 2010 to 2021 in the 10-21 folder’. This information should also be provided in the manuscript.
- In the title, I don’t understand why the term “high-resolution” is used. I would suggest to remove it.
For these reasons, the manuscript could not be published in ESSD. Authors should provide more information and validation (and cross-validation), a critical anlysis of the time series provided, which might require too much work for a major revision. I therefore concur with the recommendations from reviewers 1 and 2.
Reference:
Luo S., C. Song, P. Zhan, K. Liu, T. Chen, W. Li, and L. Ke (2021). Refined estimation of lake water level and storage changes on the Tibetan Plateau from ICESat/ICESat-2. Catena, 200, 105177, https://doi.org/10.1016/j.catena.2021.105177

Specific comments:
There is an issue concerning the way ICESat-1 is presented in the text (firstly mentioned in the text at p.2 l.48). It is never mentioned that it is a lidar altimeter and not a nadir radar altimeter, like other satellite missions mentioned in the manuscript.
p.2 l.61-62: why not adding Icesat-2, as it covers the last part of the studied time span (2002-2021)?
p.3 l.83-84: Jason-1/2/3 were not only CNES mission, please edit (see for example https://www.eoportal.org/satellite-missions/jason-1, https://www.eoportal.org/satellite-missions/jason-2, and https://www.eoportal.org/satellite-missions/jason-3). These three missions have the same orbit, so why they do not observe to the number of lakes ? Besides, Jason-1 is well known for not providing much data over continents (e.g. see at the end of section 14.2.1 in Cretaux et al., 2017). Why do you have more data from Jason-1 than from Jason-3? I am very surprised by the low number of lakes observed with Jason-3 (and the high number observed with Jason-1).
Reference:
Cretaux, J.-F., K. Nielsen, F. Frappart, F. Papa, S. Calmant, J. Benveniste (2017). Hydrological applications of satellite altimetry: rivers, lakes, man-made reservoirs, inundated areas. In: Stammer, D., Cazenave, A. (Eds.), Satellite Altimetry Over Oceans and Land Surfaces, Earth Observation of Global Changes. CRC Press, 2017.
Table 1 lacks some important information or have dubious information:
- You should add a column with the orbit repeat cycle for each mission (i.e. time sampling).
- The column with the duration is misleading. For example, if Envisat mission lasts from 2002 to 2012, in October 2010, its orbit changed (see https://www.aviso.altimetry.fr/en/missions/missions-passees/envisat.html) and therefore did its ground tracks and repeat cycle. The same goes for most missions cited in the table. Such information should be provided in the table or in the text. Besides, you should provide the time span over which you used the data. For example, for Envisat, you could only have used data from 2002 to 2010.
- I have some doubts concerning the diameter footprint provided in the last column of the table. For example, the AltiKa antenna footprint on the ground could be considered to be ~4 km (considering it corresponds to the 3-dB aperture angle of 0.6°) according to Steunou et al. (2015). Footprints of Envisat, and Jason-1/2/3 altimeters, because of Ku-band used is even coarser. Please edit the table and provide the references you used to derive information provided in Table 1.
Reference:
Steunou N., J.-D. Desjonqueres, N. Picot, P. Sengenes, J. Noubel, and J.C. Poisson (2015). AltiKa altimeter: instrument description and in flight performance. Mar. Geodesy 38 (sup1), 22–42. http://dx.doi.org/10.1080/01490419.2014.988835
Table 2: What does the column ‘Reference’ and ‘Mode’ mean and correspond to? You should add a column with the yearly mean area (or something similar) of each lake.
You should provide a map with lakes sampled with altimeters presented in section 2.2.1 and validation in situ data presented in section 2.2.2.
Section 3.1: What is the ‘satellite centroid correction’? Why do you need to consider ‘ocean tide corrections’ for lakes that are among the highest on Earth?
p.4 lines 110-120 are almost a copy/paste of Chen and Liao (2020)
p.4 l.122 and following references to Chen et al. (2020): Chen et al. (2020) is not provided in the References section.
Section 3.1.1 assumes readers are familiar with the AMPDR retracker, which is not a commonly used retrakcer. So, more information on this retracker is needed and you should define the acronyms and variables mentioned (like HDEM, DistanceThresh…). As written in lines 124 to 126, it seems that you don’t have any bias using AMPDR with Jason-2/3, Sentinel-3A/B and Cryosat-2, contrarily to other missions. I have observed bias between J2/3 and S3/B over many lakes and rivers using OCOG retracker. I am therefore more surprised that you don’t have bias with these missions, rather than you observe bias with other missions (and according to section 3.2, it seems that you have biases between each consecutive mission).
p.6 l.162: How the training set of 300 waveforms have been selected? What type of lakes do they cover?
p.6 l.167: Why excluding tracks with fewer than 5 observations?
Section 3.1.3, I don’t see what is the purpose of using tsHydro (equations 2 and 3). You should explain in more details what is the purpose of this tool, what it is supposed to do, why it is needed and all the parameters used (sigmaRW is not explained for example), why specifically equations 2 and 3 have been selected (why this model has been chosen). Providing this information in the manuscript is important, even if they are present in Nielsen et al. (2015).
p.7 l.182: I guess the ‘reference plane’ is the geoid or ellipsoid used to reference water. It would be good to provide explicitly the definition in the text.
Section 4.2 and Figure 6: Why comparing trend and not directly water elevation time series? I would prefer to see a comparison between time series. There are other altimetry water elevation databases like DAHITI (https://dahiti.dgfi.tum.de) or G-REALM (https://ipad.fas.usda.gov/cropexplorer/global_reservoir/ and https://blueice.gsfc.nasa.gov/gwm/lake/Index). You should also add these databases in your comparison
Section 5.2 is too qualitative, some assertions are not really supported by the figure (discharge seems to raise before the lake level, which is not coherent with the fact that discharge is regulated by the lakes; no information on the used precipitation is provided; there is some connection between precipitation and lake level variation and discharge; discharge time series might not be fully validated, given its ‘stair steps’ shape over some periods) and does not provide anything to your database. I suggest to delete this section.
p.15. l.320: Why does your database could be labeled as ‘high-resolution datasets’?
I download the full database and plotted all time series. Some time series looks really good, but some other raise some questions. I have not been able to load the plot on the server, but the following time series illustrates some of my observations:
- Buergacuo_Lake_Water Level.txt has a clear different behavior before and after 2011. Is it realistic?
- Co_Ngoin1_Water Level.txt has a clear annual cycle after 2011, which is not the case before and it seems strange to me.
- Kusai_Lake_Water Level.txt has an 8m water elevation increase in less than a month. Could it be due to an intermission bias?
- Laorite_Co_Water Level.txt, for some time series beofre 2011 when there are few points, time series seems to have a smooth curvy shape, which is not the case for time period with more data. Could it be due to a smoothing form the tsHydro processing?
- Gyesar_Co_Water Level.txt has too few measurements, is it worthwhile to provide this time series and compute trend?
- Xiaoquan_Lake_Water Level.txt has a 80m increase for few time steps, which does not seem realistic

Citation: https://doi.org/10.5194/essd-2022-313-RC4
- AC5: 'Reply on RC4', Jingjuan Liao, 20 May 2023
  
  General comment:
  The manuscript is easy to read and the methodology to derive water elevation is adapted, even if more information is needed (see specific comments below). However, I have the following main concerns:
  - The database is validated only for 8 lakes over 362. No information on the area of these lakes are provided, nor their locations, nor their hydrological regime. So, it’s not easy to know if they are representative of many lakes within the 362 lakes in the database.
  Reply: Thanks for the suggestions, the area, and locations of the lakes are included in the dataset. We will add an appendix table for including all the basic information (locations, areas, and also the hydrological regime info.
  - For the remaining 354 lakes, no regional consistency analysis is done between neighbor lakes. It could be a way of cross-validating the database. Similarly, for each lake, no consistency between time series from different missions is done. It would also be a way to detect if some time series might be erroneous or not compared to measurements from other missions. The intermission bias is a good way also to check if the different missions are observing the same target (if one mission provide water elevations multiples decameters above/below other missions, then they are not observing the same target).
  Reply: Thanks for the suggestions, the regional consistency analysis is a very useful tool for this dataset, at least we don’t have so many gauge stations to evaluate. We will add the cross-validating between neighbor lakes in discussion with a new section. But we don’t think this could check whether the different missions are observing the same target. There exists a system bias between different missions.
  - The comparison with other altimetry database is pretty weak. Lake trend is compared between the proposed database and the Hydroweb database. Why not comparing directly water elevations? Other altimetry databases (like DAHITI and G-REALM) should also be considered.
  Reply: Thanks for the suggestions, we will add the comparison with DAHITI, G-REALM, and HydroSat, and also direct with water elevation changes. Our initial purpose in comparing lake trends is because of various products have different errors. But the interannual trends obtained from these products should exhibit consistency. From your suggestion, we will also add the comparison with water elevation changes.
  - Multiple satellite missions are used, which have different type of sensors. For example, ICESat is a lidar altimeter, which is quite different from nadir radar altimeters (both in vertical accuracy, sensitivity to clouds…). It is never explained, nor discussed. Pros and cons from lidar and nadir radar altimeters should be provided. For example, I would expect more accurate, but less measurements in time, from ICESat than from nadir radar altimeters.
  Reply: Thanks for the suggestions. The different correction for ICEsat and nadir radar altimetry is in 3.1 section, ICEsat should consider the saturation correction, and radar altimetry should consider retracking correction. In this paper, our main goal is to generate a dataset, the different between lidar and nadir radar is not an important part. But we will still add some information to the discussion together with cross-validating between neighbor lakes.
  - Besides, there are some errors from nadir radar altimeters that are not discussed, but could have a huge impact on the database. This type of altimeters, in closed-loop tracking mode or with erroneous onboard DEM value in open-loop tracking mode, could lock their tracking window on the top of the surrounding topography near the lakes. If this topography is quite high compared to the lake (>tracking window size), the waveform will not sample the lake surface elevation. So, no matter how efficient is the retracking algorithm, it is not possible to retrieve the lake surface elevation. This point should be discussed in the manuscript. In addition to this type of error, when the lake is frozen, nadir radar altimeter could provide erroneous data. How is it delat with in the database?
  Reply: Thanks for the suggestions. The OLTC was also considered in this study when we try to retrack the waveform. Due to the influence OLTC, the waveform will show without any peaks (just like the noise signal), this could be distinguished by using the data quality flag and waveform classification. When the lake is frozen, this will be processed using retracking, actually AMPDR has the ability to obtain the water level in winter which has already be discussed in Chen et al., 2020b.
  - There are more and more published papers using satellite lidar data (ICESat, ICESat-2 and GEDI), see for example Luo et al. (2021) who studied 221 lakes in the QTP. How does your database compare to these studies?
  Reply: Thanks for the suggestions. Comparing with other published papers sound like a good situation, but we suggest to compare with HydroSat, Luo et al. (2021) has a gap losing 7 years. And our dataset is mainly using nadir radar altimetry.
  - Very few lake water elevation time series are presented (only three time series are shown, respectively in Figure 2, 3 and 4). They have rather a poor temporal sampling (figure 4) or strange elevation dynamic (figure 2).
  Reply: Thanks for the suggestions. This should be the figure looks not good, we have already revised the figure to make it looks good. About the temporal sampling, this is the normal situation because some small lakes can only be monitored by the geodetic mission Cryosat-2 without enough points, but we still add this inside, because these small lakes are also very useful for analyzing, which has been proven in Chen et al., 2020a.
  - At the database repository, it is written that ‘196 lakes have the series from 2002 to 2021 in the 02-10 folder and 168 lakes have the series from 2010 to 2021 in the 10-21 folder’. This information should also be provided in the manuscript.
  Reply: Thanks for the suggestions. This has already been included in the revised version.
  - In the title, I don’t understand why the term “high-resolution” is used. I would suggest to remove it.
  Reply: Thanks for the suggestions. We will delete the high-resolution in title.
  For these reasons, the manuscript could not be published in ESSD. Authors should provide more information and validation (and cross-validation), a critical anlysis of the time series provided, which might require too much work for a major revision. I therefore concur with the recommendations from reviewers 1 and 2.
  Reference:
  Luo S., C. Song, P. Zhan, K. Liu, T. Chen, W. Li, and L. Ke (2021). Refined estimation of lake water level and storage changes on the Tibetan Plateau from ICESat/ICESat-2. Catena, 200, 105177, https://doi.org/10.1016/j.catena.2021.105177
  Specific comments:
  There is an issue concerning the way ICESat-1 is presented in the text (firstly mentioned in the text at p.2 l.48). It is never mentioned that it is a lidar altimeter and not a nadir radar altimeter, like other satellite missions mentioned in the manuscript.
  Reply: Thanks for the suggestions. The related information of ICEsat-1 has already been added in the revised version.
  p.2 l.61-62: why not adding Icesat-2, as it covers the last part of the studied time span (2002-2021)?
  Reply: Our aim was to generate a long time series dataset, and Icesat-2 was not considered since the time period covered by Icesat-2 was also covered by Sentinel-3. We will increase the use of Icesat-2 in subsequent studies.
  p.3 l.83-84: Jason-1/2/3 were not only CNES mission, please edit (see for example https://www.eoportal.org/satellite-missions/jason-1, https://www.eoportal.org/satellite-missions/jason-2, and https://www.eoportal.org/satellite-missions/jason-3). These three missions have the same orbit, so why they do not observe to the number of lakes ? Besides, Jason-1 is well known for not providing much data over continents (e.g. see at the end of section 14.2.1 in Cretaux et al., 2017). Why do you have more data from Jason-1 than from Jason-3? I am very surprised by the low number of lakes observed with Jason-3 (and the high number observed with Jason-1).
  Reply: Thanks for the suggestions. Jason-1 will not provide much useful data in inland part, but this should be compared with Jason-2, because Jason-1 and Jason-2 all experience an interleaved orbit, increasing the number of observed lakes. While Jason-3 does not experience an interleaved orbit during 2016-2021.
  Reference:
  Cretaux, J.-F., K. Nielsen, F. Frappart, F. Papa, S. Calmant, J. Benveniste (2017). Hydrological applications of satellite altimetry: rivers, lakes, man-made reservoirs, inundated areas. In: Stammer, D., Cazenave, A. (Eds.), Satellite Altimetry Over Oceans and Land Surfaces, Earth Observation of Global Changes. CRC Press, 2017.
  Table 1 lacks some important information or have dubious information:
  - You should add a column with the orbit repeat cycle for each mission (i.e. time sampling).
  - The column with the duration is misleading. For example, if Envisat mission lasts from 2002 to 2012, in October 2010, its orbit changed (see https://www.aviso.altimetry.fr/en/missions/missions-passees/envisat.html) and therefore did its ground tracks and repeat cycle. The same goes for most missions cited in the table. Such information should be provided in the table or in the text. Besides, you should provide the time span over which you used the data. For example, for Envisat, you could only have used data from 2002 to 2010.
  - I have some doubts concerning the diameter footprint provided in the last column of the table. For example, the AltiKa antenna footprint on the ground could be considered to be ~4 km (considering it corresponds to the 3-dB aperture angle of 0.6°) according to Steunou et al. (2015). Footprints of Envisat, and Jason-1/2/3 altimeters, because of Ku-band used is even coarser. Please edit the table and provide the references you used to derive information provided in Table 1.
  Reply: Thanks for the suggestions. The orbit repeat cycle for each mission has already been added to the table. For the duration of the mission, we revised it according to different orbit, such as Jason-1/2/3 reference orbit, Jason-1/2 interleaved orbit (2009.02-2012.03 for Jason1, 2016.10-2017.05 for Jason2), Envisat reference orbit (2002.05-2010.10), and Envisat extension orbit (2010.10-2012.04). The AltiKa antenna footprint has already been revised.
  
  Reference:
  Steunou N., J.-D. Desjonqueres, N. Picot, P. Sengenes, J. Noubel, and J.C. Poisson (2015). AltiKa altimeter: instrument description and in flight performance. Mar. Geodesy 38 (sup1), 22–42. http://dx.doi.org/10.1080/01490419.2014.988835
  Table 2: What does the column ‘Reference’ and ‘Mode’ mean and correspond to? You should add a column with the yearly mean area (or something similar) of each lake.
  Reply: Thanks for the suggestions. Reference means the geoid reference, Mode corresponds to the water level is a water level or water level anomaly. We has add the table footnotes to make it easier to follow. The yearly mean area we will add all lakes’ area in the appendix table.
  
  You should provide a map with lakes sampled with altimeters presented in section 2.2.1 and validation in situ data presented in section 2.2.2.
  Reply: Thanks for the suggestions. The figure has already drawn with the location of the in situ data using the star symbol, and also the altimeters overpass orbit are also included, the satellite image map is regarded as the background.
  
  Section 3.1: What is the ‘satellite centroid correction’? Why do you need to consider ‘ocean tide corrections’ for lakes that are among the highest on Earth?
  Reply: Thanks for the suggestions. The ‘satellite centroid correction’ has already been removed, because this correction has already been added in L1b processing chain. Actually, the ocean tide corrections are all zero in Tibetan Plateau lakes, we mentioned it here is for making this formula more complete.
  
  p.4 lines 110-120 are almost a copy/paste of Chen and Liao (2020)
  Reply: Thank for your suggestions. We will modify this part.
  p.4 l.122 and following references to Chen et al. (2020): Chen et al. (2020) is not provided in the References section.
  Reply: Thanks for the suggestions. This reference has been added.
  Chen, J., Liao, J., Wang, C. (2020). Improved lake level estimation from radar altimeter using an automatic multiscale-based peak detection retracker[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 14: 1246-1259. doi.org/10.1109/JSTARS.2020.3035686.
  Section 3.1.1 assumes readers are familiar with the AMPDR retracker, which is not a commonly used retrakcer. So, more information on this retracker is needed and you should define the acronyms and variables mentioned (like HDEM, DistanceThresh…). As written in lines 124 to 126, it seems that you don’t have any bias using AMPDR with Jason-2/3, Sentinel-3A/B and Cryosat-2, contrarily to other missions. I have observed bias between J2/3 and S3/B over many lakes and rivers using OCOG retracker. I am therefore more surprised that you don’t have bias with these missions, rather than you observe bias with other missions (and according to section 3.2, it seems that you have biases between each consecutive mission).
  Reply: Thanks for the suggestions. Actually, this is a dataset description, so we don’t include too many retracking methods, as it was already published. But we also added the acronyms and variables mentioned to support this method can be easy to follow.
  AMPDR is the retracking suitable for Jason-2/3, Sentinel-3A/B and Cryosat-2, both these four satellites can get good results, which has been proved in Chen et al. (2020). This is related to the OLTC and SAR technique but still has some observations not good resulting in the bias you see, this will be finally removed from the processing of merging multi-altimetry. But for the other missions, the data quality is not so much good in inland, so we make some improve for AMPDR to make the statistics threshold more reasonable.
  p.6 l.162: How the training set of 300 waveforms have been selected? What type of lakes do they cover?
  Reply: Thanks for the suggestions. Waveforms are uniformly selected with various types of lakes, and also various locations of the lakes. Not for some type of lakes.
  p.6 l.167: Why excluding tracks with fewer than 5 observations?
  Reply: Thanks for the suggestions. This is coming from the AMPDR, this retracker are considering the statistics of the along-track water level, fewer than 5 observations will not be enough for giving stable statistics and also mean the observation is easier be polluted at terrain signals.
  Section 3.1.3, I don’t see what is the purpose of using tsHydro (equations 2 and 3). You should explain in more details what is the purpose of this tool, what it is supposed to do, why it is needed and all the parameters used (sigmaRW is not explained for example), why specifically equations 2 and 3 have been selected (why this model has been chosen). Providing this information in the manuscript is important, even if they are present in Nielsen et al. (2015).
  Reply: Thanks for the suggestions. We have added the explanation for the tshydro, such as the dynamic model formulation and its explanation.
  
  p.7 l.182: I guess the ‘reference plane’ is the geoid or ellipsoid used to reference water. It would be good to provide explicitly the definition in the text.
  Reply: Thanks for the suggestions. Yes, the ‘reference plane’ is geoid, we has added it in revised version.
  
  Section 4.2 and Figure 6: Why comparing trend and not directly water elevation time series? I would prefer to see a comparison between time series. There are other altimetry water elevation databases like DAHITI (https://dahiti.dgfi.tum.de) or G-REALM (https://ipad.fas.usda.gov/cropexplorer/global_reservoir/ and https://blueice.gsfc.nasa.gov/gwm/lake/Index). You should also add these databases in your comparison
  Reply: Thanks for the suggestions, we will add the comparison with DAHITI, G-REALM, and HydroSat, and also direct with water elevation changes. Our initial purpose in comparing lake trends is because of various products have different errors. But the interannual trends obtained from these products should exhibit consistency. From your suggestion, we will also include the comparison with water elevation changes.
  
  Section 5.2 is too qualitative, some assertions are not really supported by the figure (discharge seems to raise before the lake level, which is not coherent with the fact that discharge is regulated by the lakes; no information on the used precipitation is provided; there is some connection between precipitation and lake level variation and discharge; discharge time series might not be fully validated, given its ‘stair steps’ shape over some periods) and does not provide anything to your database. I suggest to delete this section.
  Reply: Section 5.2 focuses on a case of lake level application, the purpose of which is to show that lake level changes can reflect river regulation. Without the influence of rainfall, the water level changes of the two lakes are consistent with the discharge changes along the Yellow River, while the downstream rivers, which are not subject to the regulation of the lakes, have uneven discharge changes. Since the discharge data from gauge stations were applied, mainly qualitative relationships were derived in this study.
  
  p.15. l.320: Why does your database could be labeled as ‘high-resolution datasets’?
  Reply: Thanks for the suggestions, the high-resolution datasets is for spatial, but maybe not unclear here, we will delete it.
  
  I download the full database and plotted all time series. Some time series looks really good, but some other raise some questions. I have not been able to load the plot on the server, but the following time series illustrates some of my observations:
  - Buergacuo_Lake_Water Level.txt has a clear different behavior before and after 2011. Is it realistic?
  Reply: The lake was originally recharged by a spring. It can only be inferred that the water level tends to rise steadily after 2011 because the groundwater recharge from glacial melt is greater than the spring recharge after the temperature rises.
  
  - Co_Ngoin1_Water Level.txt has a clear annual cycle after 2011, which is not the case before and it seems strange to me.
  Reply: Thanks for the suggestions, before 2011, it is very hard to see this situation because only ICEsat can observe it. But after 2011, Cryosat-2 and Sentinel-3 could.
  
  - Kusai_Lake_Water Level.txt has an 8m water elevation increase in less than a month. Could it be due to an intermission bias?
  Reply: Thanks for the suggestions, this is not the intermission bias. Kusai Lake experienced an abrupt expansion in 2011, resulting from the dike break of an upstream lake, named Lake Zhuonai. Li et al. 2019 also reported this situation.
  
  - Laorite_Co_Water Level.txt, for some time series beofre 2011 when there are few points, time series seems to have a smooth curvy shape, which is not the case for time period with more data. Could it be due to a smoothing form the tsHydro processing?
  Reply: Thanks for the suggestions, but we think this is not the case, before 2011 looks smooth mainly because the tempoal resolution is not enough for ICEsat overpassing this lake.
  
  - Gyesar_Co_Water Level.txt has too few measurements, is it worthwhile to provide this time series and compute trend?
  Reply: Thanks for the suggestions, although is has a few point, but still can be very useful for estimating the lake level annual change rate.
  
  - Xiaoquan_Lake_Water Level.txt has a 80m increase for few time steps, which does not seem realistic
  Reply: Thanks for the suggestions, this lake has some problems when generating time series, some big bias observations was considering, we will consider remove this lakes.
  
  Citation: https://doi.org/10.5194/essd-2022-313-AC5

Jiaming Chen, Jingjuan Liao, Yanhan Lou, Shanmu Ma, Guozhuang Shen, and Lianchong Zhang

Data sets

High-space-coverage lake level change data sets on the Tibetan Plateau from 2002 to 2021 using multiple altimeter data Chen, J., Liao, J., Deng, W., Shen, G., Zhang, L., Wang, C. https://doi.org/10.1594/PANGAEA.939427

Jiaming Chen, Jingjuan Liao, Yanhan Lou, Shanmu Ma, Guozhuang Shen, and Lianchong Zhang

Viewed

Total article views: 1,953 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
1,433	447	73	1,953	68	75

HTML: 1,433
PDF: 447
XML: 73
Total: 1,953
BibTeX: 68
EndNote: 75

Views and downloads (calculated since 18 Oct 2022)

Month	HTML	PDF	XML	Total
Oct 2022	156	37	6	199
Nov 2022	123	26	5	154
Dec 2022	94	32	7	133
Jan 2023	91	22	8	121
Feb 2023	63	23	0	86
Mar 2023	46	12	1	59
Apr 2023	50	15	2	67
May 2023	59	7	2	68
Jun 2023	47	5	2	54
Jul 2023	35	24	0	59
Aug 2023	55	20	0	75
Sep 2023	68	24	2	94
Oct 2023	48	17	1	66
Nov 2023	24	13	0	37
Dec 2023	26	6	1	33
Jan 2024	39	7	1	47
Feb 2024	31	14	2	47
Mar 2024	30	24	7	61
Apr 2024	28	3	4	35
May 2024	28	9	5	42
Jun 2024	36	3	1	40
Jul 2024	19	6	3	28
Aug 2024	33	7	3	43
Sep 2024	12	10	0	22
Oct 2024	24	4	0	28
Nov 2024	26	4	0	30
Dec 2024	27	11	0	38
Jan 2025	38	20	3	61
Feb 2025	22	9	1	32
Mar 2025	18	12	4	34
Apr 2025	24	13	2	39
May 2025	13	8	0	21

Cumulative views and downloads (calculated since 18 Oct 2022)

Month	HTML	PDF	XML	Total
Oct 2022	156	37	6	199
Nov 2022	123	26	5	154
Dec 2022	94	32	7	133
Jan 2023	91	22	8	121
Feb 2023	63	23	0	86
Mar 2023	46	12	1	59
Apr 2023	50	15	2	67
May 2023	59	7	2	68
Jun 2023	47	5	2	54
Jul 2023	35	24	0	59
Aug 2023	55	20	0	75
Sep 2023	68	24	2	94
Oct 2023	48	17	1	66
Nov 2023	24	13	0	37
Dec 2023	26	6	1	33
Jan 2024	39	7	1	47
Feb 2024	31	14	2	47
Mar 2024	30	24	7	61
Apr 2024	28	3	4	35
May 2024	28	9	5	42
Jun 2024	36	3	1	40
Jul 2024	19	6	3	28
Aug 2024	33	7	3	43
Sep 2024	12	10	0	22
Oct 2024	24	4	0	28
Nov 2024	26	4	0	30
Dec 2024	27	11	0	38
Jan 2025	38	20	3	61
Feb 2025	22	9	1	32
Mar 2025	18	12	4	34
Apr 2025	24	13	2	39
May 2025	13	8	0	21

Viewed (geographical distribution)

Total article views: 1,852 (including HTML, PDF, and XML) Thereof 1,852 with geography defined and 0 with unknown origin.

Country	#	Views	%

Cited

Latest update: 18 May 2025

Short summary

The Qinghai Tibet Plateau (QTP) has the largest number of lakes in the world. Hydrological stations cannot be readily set up in this region due to high altitude and remoteness. Satellite radar altimetry has become a very important alternative to in situ observations as a source of data. By combining eight sets of altimeter data, the trends of the changes in the water levels for 362 lakes in the QTP during 2002–2021 were estimated, and the spatio-temporal changes in lake levels were explored.


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

High-resolution datasets for lake level changes in the Qinghai-Tibetan Plateau from 2002 to 2021 using multi-altimeter data

Data sets

Viewed

Viewed (geographical distribution)

Cited

5 citations as recorded by crossref.