A global seamless 1&thinsp;km resolution daily land surface temperature dataset (2003–2020)

Zhang, Tao; Zhou, Yuyu; Zhu, Zhengyuan; Li, Xiaoma; Asrar, Ghassem R.

doi:https://doi.org/10.5194/essd-14-651-2022

Articles | Volume 14, issue 2

https://doi.org/10.5194/essd-14-651-2022

© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/essd-14-651-2022

© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 14, issue 2

Data description paper

|

15 Feb 2022

Data description paper |

| 15 Feb 2022

A global seamless 1 km resolution daily land surface temperature dataset (2003–2020)

Tao Zhang, Yuyu Zhou, Zhengyuan Zhu, Xiaoma Li, and Ghassem R. Asrar

Download

Final revised paper (published on 15 Feb 2022)
Supplement to the final revised paper
Preprint (discussion started on 12 Oct 2021)
Supplement to the preprint

Interactive discussion

Status: closed

RC1: 'Comment on essd-2021-313', Anonymous Referee #1, 01 Nov 2021

Comments to the authors

This paper proposes a very interesting spatiotemporal gap-filling framework to generate seamless global 1 km daily (mid-daytime and mid-nighttime) LST products from 2003 to 2020. The manuscript is clearly written and well-structured. The products provide global coverage and better accuracies for long-term time-series LSTs. After incorporating my suggested edits, I recommend acceptance.

Major comments

1. The novelty of this work is not comprehensively presented throughout the study. The global significance and necessity of this work should be clarified further. It can be more clearly presented in the abstract, introduction, and conclusion sections.

2. A more detailed explanation of the methodology is necessary. I suggest that SI and S2 should be moved to Section 3.2 (spatiotemporal fitting).

3. More comparisons to previous methods are necessary for the discussion section to show the validity of your method, except for Li et al. (2018).

4. The method may help reconstruct seamless global daily LST products based on MODIS datasets, but the study is limited on clear-sky conditions. How about the LSTs under cloudy conditions? More discussions may be needed.

Minor Comments

1. Line 21 and Line 237-242: Compared with other gap-filling methods, further details should be provided as to how to quantify the effectiveness and efficiency of the gap-filling framework.

2. Line 94-95: confusing sentence, please rephrase.

3. Line 111: What are the criteria for identifying the PVD at 5%.

4. I understand Section 3.1 is from previous studies, but it is necessary to provide more basic information about this section.

5. Fig 2 should be revised.

6. Put more details for Fig 3.

7. What are the advantages of using the smoothing spline function?

8. I did not see any consideration about PVD<5% of four observations from the two satellites.

9. Line 138-139: Not clear of how you manually introduced under three scenarios gaps here? Are the three gaps randomly distributed or continuously missing? How was the reconstruction accuracy in different distributions?

10. Why does the number of pixels (N range from 9300 to 9350) vary approximately among different excluded areas (25%, 50%, and 75%)?

11. Line 155: I consider the selection of 10 pixels per image seems to be not reasonable for cross-validation analysis.

12. Line 214 does not provide a citation?

13. Line 219-224: I saw this very similar sentence in the introduction.

Citation: https://doi.org/10.5194/essd-2021-313-RC1
RC2: 'Comment on essd-2021-313', Anonymous Referee #2, 02 Nov 2021

Zhang et al. proposed a novel method to generate a seamless global 1 km daily MODIS-like LST dataset. The topic is very interesting and the performance of the method looks good. The newly developed dataset should be of great use in climatic and ecological studies. I would like recommend acceptance after following minor revisions.

Line 15: “first, we fitted the long-term trend (overall mean) of observations in each pixel (ordered by day of year)” is confusing. First, the “overall mean” is ambiguous. Second, the independent variable should be included (e.g., Day of year?). Last, the fitting method should be introduced as well.

Lines 40-42: Some descriptions are not so accurate. For example, the spatial resolution of the Landsat series and ASTER data should be 60-120m; the temporal resolution for the ASTER data is not 16 days; the MODIS data should not be termed as “high spatial resolution about 1 km and high temporal resolution of….” as compared to the former two groups of the datasets. May be “moderate” is more appropriate.

Line 57: It remains unclear what are the so-called “there are still limitations in these products” here. It seems that this paragraph list some previous studies without pointing out the specific knowledge gaps. Maybe it’s better to incorporate this paragraph into the following one.

Line 75: I think the method based on empirical relationship is not necessarily computationally expensive as compared to the spatiotemporal interpolation and hybrid methods. For example, I came across studies that filled the missing LST values based on the ERA5-Land skin temperature might be more efficiently though with low accuracies.

Line 97: The legends of figure 1 are somewhat confusing. I think this study filled the gaps for all the tiles. Thus the “Title for gap-filling” should be removed to avoid possible misleading.

Line 102-108: Most descriptions here are completely the same as that in Abstract. Given that these have been detailed in the following sections, it’s better to simplify or remove the duplications.

Line 115: “If PVD of T2 >= 5%, did not change the data and accepted it, otherwise we gap-filled missing values based on the following order”: I am lost on whether this study filled the missing values or not for the “PVD of T2 >= 5%”.

Lines 116-118: Did this study fills the missing values by all the three methods or just one of them for a given pixel location?

Line 118: A brief introduction to the “shift methods” is needed here.

Line 123: Does “the overall mean of observations in each pixel” mean the pixel with all the four valid observations?

Lines 127-128: I have two confusions here. First, it remains unclear which type (e.g., linear regression or others) of the correlation has been used and how many of the neighboring valid pixels (or how large the neighborhood size) have been used for each target pixel. Second, if the value for the target pixel is missing, how did this study obtain the correlation between the target pixel and its neighboring valid pixels?

Line 137: “In each of these years, we selected 19 days with the largest coverage of high quality data (coverage > 95% quantile)”: It remains unclear what does the coverage mean? the coverage of all the four observations or the “overall mean” in a day ?

Lines 160: Besides the contrasting performances between urban and rural areas, the accuracies may vary substantially by the climate zones due to the different data availability and seasonality. For example, the tropical regions may have lower accuracies than the high latitude regions. I am wondering if possible to show the model performances for difference climate zones.

Line 179: I think we cannot see the UHI phenomenon from figure 6 if we do not know where the urban areas are.

Line 179: It would be much better if this study can summarize the existing seamless LST data by a table.

Citation: https://doi.org/10.5194/essd-2021-313-RC2
CC1: 'Comment on essd-2021-313', Aolin Jia, 25 Nov 2021

Dear Dr. Zhou,

It’s a great pleasure for me to participate in this open discussion. This paper proposed clearly significant work for improving the practicality of the current MODIS LST product. As my phD dissertation is also related to cloudy-sky LST estimation and LST data applications, this work highly attracts me and I believe if the released dataset achieves satisfactory accuracy and data quality, it will definitely help the communities in monitoring surface thermal dynamics, .etc.

I do have several questions and concerns that I cannot fully understand from the paper, therefore, I would like to list them below, and would you please have a clarification?

Thank you for your work and contributions to the topic. Also, I sincerely appreciate your time for explaining my questions.

Warm regards,
Aolin Jia
PhD candidate
Department of Geographical Sciences
University of Maryland, College Park, MD, USA

Major Questions:
1) the top question I would like to ask is related to the major comment (4) from Referee #1. The assessment method of the proposed work only focused on clear-sky samples, whereas recovered cloudy-sky results are the main contributions of this study that should be focused on.
This study assessed the results by hypothetically removing some clear-sky days and filling them, and then the filled results were compared with the official MODIS LST. However, not like the shortwave variables (surface reflectance), LST values under clouds are affected by the cloud coverage (cooling/cooling effect at daytime/nighttime), thus clear-sky samples are not representative for the cloudy-sky cases.
Besides, the hypothetically removed clear-sky pixels usually have enough clear-sky adjacent pixels for interpolation, and they can easily get good interpolation accuracy than actual cloudy cases, which might be not fair to only use this assessment method to demonstrate the accuracy in cloudy-sky. Moreover, this assessment method cannot involve the filled LST samples where official MODIS has bad data quality (partially cloud-covered or large view zenith angle), while they are also important parts of the contributions of this work.
Therefore, I am really curious about the accuracy validated by ground measurements that can comprehensively assess both clear-sky and cloudy-sky samples, independently. This work has mentioned that site measured record is not comparable to remote sensing retrieved LST due to spatial scale mismatch or broadband emissivity (BBE) uncertainty. In fact, those issues have been discussed in previous studies. Li et al. (2021) have proved that SURFRAD sites have little heterogeneity issue for validating 1-km LST product by analyzing Landsat LST, and BBE would not introduce noticeable errors for the validation (Xing et al., 2021). And that’s why most related studies in this topic used ground measurement for assessment (Zhang et al., 2021, Xu and Cheng, 2021, Zeng et al., 2018).
The proposed work followed Li et al. (2018) and didn’t consider the site validation, but I think this is mainly because Li et al. (2018) only focused on US urban areas that there is no available ground LST or upward longwave radiation measurements, besides, the heterogeneity issue in urban areas is substantial and cannot be ignored. As this study generated the global gap-free LST, it would be a great opportunity to assess the product by using high-quality site measurements, such as SURFRAD and BSRN, globally.
It is reasonable that the site-validated RMSE is not comparable to the cross-validation results, but by separately validating clear-sky and cloudy-sky results using ground site measurement and comparing the RMSEs in clear days and cloudy days, this study can demonstrate the accuracy stability and real interpolation accuracy of the proposed product, which I believe will be what users care about.

2) I’m still a little bit confused about the methodology. Would you please use plain language to explain how the interpolation methodology captures the day-to-day variation signals where there are cloud covers with considerable spatiotemporal scales, such as raining seasons (continuity could be weeks) in southern China or tropical areas without introducing passive microwave or modeling data?

3) Interpolation-based method usually has a trade-off between calculation efficiency and accuracy because essentially they used statistical relationships with referred pixels, the more referred pixels they can obtain, the higher accuracy they will achieve, but more time will be consumed. This category of the methodology has been well developed and would you please strengthen the breakthrough of this proposed work, and what are its differences from previous interpolation methods, and how could the efficiency be improved while high accuracy is maintained?

4) I also have a question about neighboring pixels in a spaital window: it looks like land cover type and elevation differences were not considered in the method (maybe I missed). Land cover types highly impact the LST values even they have close spatial distance. For example, forest LST is the thermal signal from the canopy, but neighboring LST of open grassland could be from the ground (even soil surface), which are significantly different. That’s why Zhang et al. (2021) build models by land cover types in each window. Jin and Treadon (2003) also build typical diurnal temperature cycles for each land cover type. May I ask if you have included such consideration? If not, could you please have some discussion to clarify that those differences won’t significantly affect the interpolation accuracy?

Other Comments:
I also have some minor comments that might be helpful for the paper publication.
1) I found some typos and grammar errors in the manuscript so would you please revise them? These are not all of them so I would suggest that English editing would improve the writing quality.
Line 11: parameter -> parameters
Line 35: measured is not accurate, could be retrieved
Line 40: ASTER -> Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER)
Line 47: “a” is missing before “larger number”
Line 50: “a” is missing before “daily”
Line 57: “the” is missing before conterminous US.
Line 65: relationships
Line 71: “a” is missing before hybrid method
Line 83: “the” is missing before “methods mentioned above”
Figure 1: would you please add H and V numbers to the x-axis and y-axis of Figure 1
Line 102: first “)” could be removed
Line 115: “If PVD of T2 >= 5%, did not change the data and accepted it” would you please explain this sentence? and why 5%?
Line 138: ‘the’ is missing before selected days
Line 139: ‘originally values’ -> ‘original values’
Line 158: numbers, ‘are’ -> being
Line 175: in day 200 -> on
Line 182: are -> is
Line 210: exist
Line 218: has -> have
Line 254: ‘those’ could be removed?

2) in the supplementary materials, would you please explain why this method refers to center pixels from neighboring blocks for interpolation (could be 10 km far from the target pixel) rather than using neighboring pixels. S1 is easily understood but could you have a clarification on why the “S2 Implementation of the ICW method” is designed like this? Thanks!

References
Jin M. & Treadon R. 2003. Correcting the orbit drift effect on AVHRR land surface skin temperature measurements. International Journal of Remote Sensing, 24(22), 4543-4558.
Li B., Liang S., Liu X., Ma H., Chen Y., Liang T. & He T. 2021. Estimation of all-sky 1 km land surface temperature over the conterminous United States. Remote Sensing of Environment, 266, 112707.
Li X., Zhou Y., Asrar G. R. & Zhu Z. 2018. Creating a seamless 1 km resolution daily land surface temperature dataset for urban and surrounding areas in the conterminous United States. Remote Sensing of Environment, 206, 84-97.
Xing Z., Li Z.-L., Duan S.-B., Liu X., Zheng X., Leng P., Gao M., Zhang X. & Shang G. 2021. Estimation of daily mean land surface temperature at global scale using pairs of daytime and nighttime MODIS instantaneous observations. ISPRS Journal of Photogrammetry and Remote Sensing, 178, 51-67.
Xu S. & Cheng J. 2021. A new land surface temperature fusion strategy based on cumulative distribution function matching and multiresolution Kalman filtering. Remote Sensing of Environment, 254, 112256.
Zeng C., Long D., Shen H., Wu P., Cui Y. & Hong Y. 2018. A two-step framework for reconstructing remotely sensed land surface temperatures contaminated by cloud. ISPRS Journal of Photogrammetry and Remote Sensing, 141, 30-45.
Zhang X., Zhou J., Liang S. & Wang D. 2021. A practical reanalysis data and thermal infrared remote sensing data merging (RTM) method for reconstruction of a 1-km all-weather land surface temperature. Remote Sensing of Environment, 260, 112437

Citation: https://doi.org/10.5194/essd-2021-313-CC1
AC1: 'Comment on essd-2021-313', Yuyu Zhou, 20 Dec 2021

The comment was uploaded in the form of a supplement: https://essd.copernicus.org/preprints/essd-2021-313/essd-2021-313-AC1-supplement.pdf

Citation: https://doi.org/10.5194/essd-2021-313-AC1

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

AR by Yuyu Zhou on behalf of the Authors (20 Dec 2021) Author's response Author's tracked changes Manuscript

EF by Manal Becker (22 Dec 2021) Supplement

ED: Referee Nomination & Report Request started (30 Dec 2021) by Kirsten Elger

RR by Anonymous Referee #2 (03 Jan 2022)

RR by Anonymous Referee #1 (16 Jan 2022)

ED: Publish as is (16 Jan 2022) by Kirsten Elger

AR by Yuyu Zhou on behalf of the Authors (18 Jan 2022) Author's response Manuscript

Download

Article (4849 KB)
Full-text XML

Short summary

We generated a global seamless 1 km daily (mid-daytime and mid-nighttime) land surface temperature (LST) dataset (2003–2020) using MODIS LST products by proposing a spatiotemporal gap-filling framework. The average root mean squared errors of the gap-filled LST are 1.88°C and 1.33°C, respectively, in mid-daytime and mid-nighttime. The global seamless LST dataset is unique and of great use in studies on urban systems, climate research and modeling, and terrestrial ecosystem studies.