the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 13 Jul 2026
Impact of aeolian sediments on the ice-covered lake surface temperature of Zonag Lake, Tibetan Plateau
Abstract. Aeolian sediments on ice-covered lakes can modify surface energy conditions and potentially influence the thermal environment of lake ice. However, the quantitative relationship between aeolian sediments and lake ice surface temperature remains insufficiently understood. This study investigates the relationship between aeolian sediments and ice-covered lake surface temperature (LST) in Zonag Lake on the Tibetan Plateau using multi-temporal Landsat 8 OLI imagery and MOD11A1 LST data from 2013 to 2018. The Normalized Difference Sandy Land Index (NDSLI) was applied to identify the distribution of aeolian sediments on lake ice. To quantify the sediment impact on LST, spatial regression analyses were conducted using ordinary least squares (OLS), geographically weighted regression (GWR) and multi-scale geographically weighted regression (MGWR). The results show that NDSLI effectively captures the spatial distribution and migration patterns of aeolian sediments on lake ice. Sediment-covered areas (NDSLI < 0) consistently exhibit higher temperature than uncovered areas (NDSLI > 0), with observed temperature differences ranging from 0.83 to 2.36 °C. Finer-scale NDSLI classification indicates a progressive increase in surface temperature with increasing sediment accumulation. Spatial regression results reveal a significant positive relationship between sediment and LST, with MGWR providing the best model performance and explanatory capability among the tested models. This study provides quantitative evidence that aeolian sediment deposition can modify the thermal conditions of lake ice surfaces, offering new insights into cryosphere–aeolian interactions in arid-cold regions and improving the understanding of thermal variability in ice-covered lakes.
- Preprint
(1727 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 18 Oct 2026)
-
CC1: 'Comment on egusphere-2026-2129', Lei Zheng, 24 Jul 2026
reply
-
AC1: 'Reply on CC1', Guangyin Hu, 18 Aug 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-2129/egusphere-2026-2129-AC1-supplement.pdf
-
AC1: 'Reply on CC1', Guangyin Hu, 18 Aug 2026
reply
-
CC2: 'Comment on egusphere-2026-2129', Ke Zhang, 01 Aug 2026
reply
This is a well-conducted and timely study that provides quantitative evidence for the thermal perturbation of ice-covered lake surfaces by aeolian sediments on the Tibetan Plateau. The application of spatial regression models, particularly MGWR, to quantify sediment-induced thermal anomalies is appropriate and helps address an important knowledge gap regarding cryosphere–aeolian interactions. The results indicate that sediment-covered areas exhibit surface temperature increases of 0.83–2.36°C, highlighting the potential role of aeolian sediments as an important non-climatic driver of lake ice thermal heterogeneity. The manuscript is generally well organized, and the methodology appears robust. I have a few comments that I believe will further strengthen the manuscript.
Specific comments
1. The Landsat 8 and MODIS datasets are not always acquired on the same date. Given that aeolian sediment streaks can migrate rapidly under strong wind conditions, this temporal mismatch may introduce uncertainty into the spatial correlation analysis. The authors should discuss the likely migration rate or persistence of these sediment streaks and evaluate how such temporal offsets may affect the reported relationships. If available, daily meteorological observations from nearby weather stations could also be used to demonstrate that background thermal conditions remained relatively stable during the interval between the two satellite acquisitions, thereby helping to constrain this source of uncertainty.
2. The 2011 Zonag Lake outburst and the subsequent reorganization of the regional lake system, which resulted in extensive exposure of former lakebeds, represent a major environmental change in the study area. Although this event is briefly introduced in the study area description, its potential implications are not fully explored in the subsequent analysis. Since the newly exposed lakebed likely became the dominant source of aeolian sediments, it would be informative to examine whether the observed sediment–temperature relationship became stronger or more evident after 2011 compared with earlier periods. At a minimum, the authors should discuss this possibility in Section 5, as it would provide valuable environmental context for interpreting the results.
3. Line 77: A punctuation mark is needed between "conditions" and "Monitoring"because these are two independent sentences. The sentence should read:"...regional aeolian conditions. Monitoring data revealed..."
4. The first paragraph of Section 5.2 would fit better in the Introduction. It clearly outlines the scientific significance of the study and identifies the key research gap, providing a stronger rationale for the work at an earlier stage of the manuscript. Moving this paragraph to the Introduction would improve the overall flow and better frame the objectives of the study.
Citation: https://doi.org/10.5194/egusphere-2026-2129-CC2 -
AC2: 'Reply on CC2', Guangyin Hu, 18 Aug 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-2129/egusphere-2026-2129-AC2-supplement.pdf
-
AC2: 'Reply on CC2', Guangyin Hu, 18 Aug 2026
reply
-
CC3: 'Comment on egusphere-2026-2129', Ke Zhang, 01 Aug 2026
reply
This is a well-conducted and timely study that provides quantitative evidence for the thermal perturbation of ice-covered lake surfaces by aeolian sediments on the Tibetan Plateau. The application of spatial regression models, particularly MGWR, to quantify sediment-induced thermal anomalies is appropriate and helps address an important knowledge gap regarding cryosphere–aeolian interactions. The results indicate that sediment-covered areas exhibit surface temperature increases of 0.83–2.36°C, highlighting the potential role of aeolian sediments as an important non-climatic driver of lake ice thermal heterogeneity. The manuscript is generally well organized, and the methodology appears robust. I have a few comments that I believe will further strengthen the manuscript.
Specific comments
1. The Landsat 8 and MODIS datasets are not always acquired on the same date. Given that aeolian sediment streaks can migrate rapidly under strong wind conditions, this temporal mismatch may introduce uncertainty into the spatial correlation analysis. The authors should discuss the likely migration rate or persistence of these sediment streaks and evaluate how such temporal offsets may affect the reported relationships. If available, daily meteorological observations from nearby weather stations could also be used to demonstrate that background thermal conditions remained relatively stable during the interval between the two satellite acquisitions, thereby helping to constrain this source of uncertainty.
2. The 2011 Zonag Lake outburst and the subsequent reorganization of the regional lake system, which resulted in extensive exposure of former lakebeds, represent a major environmental change in the study area. Although this event is briefly introduced in the study area description, its potential implications are not fully explored in the subsequent analysis. Since the newly exposed lakebed likely became the dominant source of aeolian sediments, it would be informative to examine whether the observed sediment–temperature relationship became stronger or more evident after 2011 compared with earlier periods. At a minimum, the authors should discuss this possibility in Section 5, as it would provide valuable environmental context for interpreting the results.
3. Line 77: A punctuation mark is needed between "conditions" and "Monitoring"because these are two independent sentences. The sentence should read:"...regional aeolian conditions. Monitoring data revealed..."
4. The first paragraph of Section 5.2 would fit better in the Introduction. It clearly outlines the scientific significance of the study and identifies the key research gap, providing a stronger rationale for the work at an earlier stage of the manuscript. Moving this paragraph to the Introduction would improve the overall flow and better frame the objectives of the study.
Citation: https://doi.org/10.5194/egusphere-2026-2129-CC3 -
AC3: 'Reply on CC3 （The same with the "Reply on CC2"）', Guangyin Hu, 18 Aug 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-2129/egusphere-2026-2129-AC3-supplement.pdf
-
AC3: 'Reply on CC3 （The same with the "Reply on CC2"）', Guangyin Hu, 18 Aug 2026
reply
-
RC1: 'Comment on egusphere-2026-2129', Anonymous Referee #1, 02 Sep 2026
reply
Overall comment: The topic is interesting and the spatial patterns shown in the paper are worth investigating. My main concern is that the current analysis shows a strong spatial relationship between sediment distribution and LST, but this is sometimes presented as proof that sediments are causing the temperature differences. I think the paper will be much stronger if the authors explain the Landsat-MODIS matching clearly, provide more information about the satellite processing and regression models, address the uncertainty in the data, and make the conclusions more careful.
Major Comments
- I am concerned about how the Landsat and MODIS images were matched. Some of the images are several days apart. For example, the Landsat image from December 26, 2017 is matched with MODIS LST from December 20, 2017. There is a six-day difference. Since the paper shows that sediments can move across the lake surface with wind, how can we be sure that the sediment distribution six days later represents the conditions when the MODIS temperature was measured? Weather conditions could also change during those days. The authors should show the time difference for every Landsat-MODIS pair and explain why these differences are acceptable. If possible, the main analysis should use same-day observations.
- This part needs more explanation. Landsat NDSLI is at 30 m while MODIS LST is at 1 km. The paper says that MODIS LST was resampled to “higher resolution (1 km × 1 km),” but MOD11A1 is already 1km. How exactly were the two datasets combined? Was the 30 m NDSLI aggregated to 1 km? If so, was the mean NDSLI used, the percentage of sediment-covered pixels, or something else? If MODIS was resampled to 30 m, this would not create new temperature and could result in many Landsat pixels sharing the same MODIS temperature. The authors need to clearly explain the complete spatial matching and resampling process.
- The paper says MOD11A1 V6 was used, but important processing information is missing. Was daytime or nighttime LST used? What MODIS quality flags were applied? How were cloudy pixels removed? How were mixed shoreline pixels handled? What was the acquisition time? This is especially important because some of the reported temperature differences are less than 1 °C. The authors should also discuss the uncertainty of MODIS LST over snow- and ice-covered surfaces.
- The authors use NDSLI < 0 to identify sediments and NDSLI < -0.15 for stronger sediment accumulation. However, frozen lake surfaces can have different spectral conditions because of snow, bare ice, cracks, wet ice, snow depth, and surface roughness. How can we be sure that negative NDSLI values are caused mainly by sediments? The visual comparison in Figure 2 is useful, but I do not think it is enough to say that the sediment boundaries were “accurately extracted.” The authors should provide additional validation or test how the results change when different NDSLI thresholds are used or soften the language
- The text refers to Gao et al. (2025), including when discussing the previous validation of NDSLI. However, the reference list gives the related paper as Gao et al. (2026). Please check and make the citation year consistent.
- The authors report a 6.28 °C difference by comparing the maximum temperature in a sediment-covered area with the minimum temperature in a sediment-free area. I do not think this is the best way to estimate the sediment effect because these are two extreme values from different locations. The mean or median temperature difference between sediment and non-sediment areas is more useful. Confidence intervals or another measure of uncertainty would also help.
- Figure 7 shows that LST decreases as NDSLI increases, meaning that the relationship between NDSLI and LST is negative. Lower NDSLI means more sediment, so the relationship between sediment amount and LST is positive. These are not the same statement. The paper sometimes describes the relationship simply as a positive relationship. The authors should make this distinction clear throughout the paper.
- The paper mainly presents R², adjusted R², AICc, bandwidth, and sigma. However, the actual NDSLI regression coefficients are not shown. If the purpose of the model is to understand how NDSLI relates to LST, the coefficient is very important. Please report the coefficient values and their uncertainty or significance. For GWR and MGWR, maps showing the local coefficients would also be useful.
- The paper explains that MGWR is useful because different explanatory variables can operate at different spatial scales. However, the model appears to have only one main predictor, NDSLI. The authors should explain more clearly what MGWR adds in this particular case compared with GWR. The paper should also provide the kernel type, whether the bandwidth is fixed or adaptive, how bandwidth was selected, the distance method, and other important model settings.
- The manuscript repeatedly says that MGWR consistently performs better than GWR. However, this is not true for March 31, 2018. GWR has an adjusted R² of 0.442 and AICc of 355.202, while MGWR has an adjusted R² of 0.438 and AICc of 357.377. In this case, GWR performs slightly better. The authors should change “consistently outperformed” to something like “generally outperformed.”
- The manuscript repeatedly says that MGWR consistently performs better than GWR but that is not the case. Please change the language
- The authors do not explain how the spatial weights were created. Was queen contiguity, rook contiguity, distance, or k-nearest neighbours used? Were the weights standardized? How many permutations were used to determine significance? These details are necessary to reproduce the analysis. Also, the equations for Moran’s I should be checked carefully because Equation 2 does not appear to be written correctly.
- NDSLI appears to be the only main explanatory variable in the regression. However, lake ice temperature can also be influenced by air temperature, snow cover, radiation. Etc. The conclusion currently says that the results “robustly confirm” I think this should be changed to something more careful, such as saying that sediment coverage is strongly associated with the spatial variation in LST.
- The Methods say that images were selected when the lake was in a stable freezing period. However, the Results later say that one image represents a non-freezing condition and another represents snowfall conditions. If these images were intentionally included as comparison or control cases, please explain this clearly in the Methods.
Minor Comments
The Results move from 4.1 to 4.3, then 4.4, 4.5, and later back to 4.4. This is confusing.
The caption to Figure 8 reads OSL where OLS is intended.
L26: “Sediments accumulation” should be “sediment accumulation”. There are other places in the document with the same issue
L32: “Researches” should be “Research.”. Other parts of the document have this issue as well
L42: “Beside” should be “Besides.”
L47: “The” is missing. Change to: “At the Headwater Region of the Yellow River”
L50: Change “In Mars” to “On Mars.”
L31: “The” is missing in two places.
Change to: “...between aeolian sediments and the ice-covered period of the lake.”L77: A period is missing after “conditions.”
L189: Remove “was.” Change to: “NDSLI < 0 accounted for 46.1% of the total lake surface.”
L231: “These evidences indicate that the aeolian sediments accumulation significantly changes the thermal distribution of ice-covered lake surface...” Check this sentence
L249: “Exhibit” should be “exhibits,” and “a” is missing before “bimodal pattern.”
L267: “Only one date showed relatively good fitting result (R² = 0.567) is December 31, 2013, whereas other dates exhibited poor fitting results...” Check this sentence
L347: “Researches” should be “research.”
L408: “The authors declare that they have no conflict of interest..”
There is an extra period at the end.
There are some date inconsistencies
Table 1: Landsat 2017-11-08 / MODIS 2017-11-07
Figure 3(d): 2017-11-07
Figure 4(e): 2017-11-20
Section 4.4, Line 208: November 20, 2017The dates are not consistent. Please check whether November 20 is a typo or a different observation date that was not included in Table 1.
This is important because this date is linked to the reported Moran’s I value of 0.8327.
Citation: https://doi.org/10.5194/egusphere-2026-2129-RC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 180 | 79 | 23 | 282 | 15 | 13 |
- HTML: 180
- PDF: 79
- XML: 23
- Total: 282
- BibTeX: 15
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript addresses a timely and important topic at the intersection of aeolian processes and ice-covered lake dynamics. The study’s focus on the Tibetan Plateau is particularly relevant given the region’s sensitivity to climate change and the significant sediment sources. I would like to offer a few comments on the current version.
1. The introduction provides a broad review of light-absorbing particles and snow/ice processes, but it remains largely descriptive. The manuscript would benefit from a sharper articulation of the specific knowledge gap this study addresses. What is currently known versus unknown about the thermal effects of aeolian sediments on lake ice, and how the present work fills that gap. Strengthening this framing would better justify the novelty and significance of the research.
2. Several methodological choices require clearer justification. It would be helpful to explain why MOD11A1 was selected over other available surface temperature products, and why Landsat 8 OLI was chosen for sediment mapping given that higher-resolution alternatives (e.g., Sentinel-2) exist. The thresholds for NDSLI are introduced without quantitative validation, providing accuracy metrics based on at least manual interpretation would strengthen the sediment extraction results.
3. The advantages over conventional GWR are not clearly demonstrated. Is the multiscale GWR framework truly necessary? Moreover, the broadband albedo is used as a key argument for the proposed radiative mechanism. If this analysis is essential, the corresponding data, methods, and results should be moved to the main body.