Assessment of MODIS Surface Temperature Products of Greenland Ice Sheet Using In-Situ Measurements
, Yetang Wang 1, Weijun Sun 1, Qinglin Zhang 1 and Baojuan Huai 1,*Round 1
Reviewer 1 Report
Manuscript: land-1661228-peer-review-v1.pdf
Title: Assessment of MODIS Surface Temperature products of Greenland using In-situ Measurements
Authors: Xiaoge Yu, Tingting Wang, Minghu Ding, Yetang Wang, Weijun Sun, Qinglin Zhang and Baojuan Huai
The manuscript is essentially concerned with evaluating the accuracy of the ice surface temperature (IST) retrieved by MODIS with in-situ measurements in Greenland over the past 20 years. In addition, both temporal and spatial patterns of the parameter and its trends are shown and discussed.
Quantifying and understanding the variability of the IST (and other physical parameters) through accurate satellite measurements, is certainly useful and interesting, especially in view of the current problem of the amplified global warming at higher latitudes.
In my opinion, a weakness of the work described in the manuscript is represented by the quantitative statistical methodology used, practically limited to the calculation of certain quantities such as bias, rmse, Pearson's correlation coefficient, and slope of trends. This necessarily implies a loss in the novelty of the manuscript.
However, after some adjustments, I believe the manuscript may be publishable as is.
Hereafter my (minor) comments for each section:
[Abstract]
No comments. It looks well.
[1. Introduction]
Looking at the reference [18] quoted on page 1, I did not find the point where it is reported that “the GrIS alone could lead to 9 cm of sea level rise by 2050”. It's probably my oversight. In any case, 9 cm is a value compatible with other estimates, for example see Schwanden et al. (2019).
[2. Material and methods]
- In Sect. 2.3 I would suggest entering the information on how many pixels have been removed.
- Were the trends (their slopes) calculated with the standard parametric method (i.e., OLR)? If so, has their statistical significance (at (1-alpha)%) been estimated, and if so why has it not been shown in a table or directly in Figure 8?
[3. Results]
- In some scatterplots of Figure 2 (e. g., TAS_U and TAS_L) I seem to notice an evident non-gaussianity in the distribution of the data points, and also the presence of large outliers, which would make the calculated coefficient of determination R2 inaccurate. What do the authors answer?
- Given that, in general, it is to be avoided to connect discrete points unless there is either a scientific to the implied interpolation, I would suggest that, in figure 3, the points between a region (SE, SW, etc. ) and another were disconnected (the fourth with the fifth, the fourteenth with the fifteenth, etc.). I think it would improve the figure.
- Figure 4 should be removed because the relevant features are already clear from the body of text. Alternatively, the authors could show only the differences between the two timeseries.
- In citing [3] on page 8, I would think it would be better to specify that the estimates for the period 1960-2010 derive in part from simulated data
[4. Discussion]
- Please, if possible, enlarge Figure 9.
- Figure 10, for the same reasons expressed for Figure 3, I would avoid connecting the points, or I would suggest using a bar diagram.
- The two decimal digits reported in the percentage values on page 10 commenting on figures 9 and 10, in my opinion, should not be reported.
- Rather than the reference [55] it would seem right to quote "Van der Brooke et al, 2016" quoted by the reference [55] itself.
[5. Conclusions]
The conclusion is ok, but it is too short. I would suggest extending it or creating a single section “Discussion and Conclusion” merging sect. 4 and 5.
Typos
[Page 4, Line 10] There is something wrong: “Its algorithm derived from the algorithm ...”.
[Page 4, Line 20 and Page 9, Line 3] replace “Hall et al., 2013” with the correct reference number (I presume 5).
[Page 9, Line 18] replace ‘area’ with ‘aera’.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
land-1661228 Review
Dear Editors, Dear Authors,
The MS written by Yu et al. 2022 presents an attempt to detect a temperature of the ice (or rather snow!) on the surface of the Greenland Ice Sheet using remote sensing on the basis of data captured by the Terra and Aqua satellites launched by NASA (US federal Space Agency). The authors compare the remotely sensed results with field data gathered by 24 automatic meteorological stations run by GEUS (Geological Survey of Denmark and Greenland) and DTU Orbit (Denmark Technical Univ; National Space Ins.in Kongens Lyngby near Copenhagen) and installed on the glacial ice or above the ice in the firn or snow.
In the 1st section the Authors drew a dramatic view of melting of the 2nd largest ice-sheet on Earth and unfortunately linked it with “climate change” and “sea level rise” what I found like an “affirmation action” of controversial topic originated from Western politics.
In the 2nd section they described the data sources: (1) network of meteorological stations, (2) satellites, and (3) processing of the satellite data with specific emphasis on the (4) statistical approach, including (5) anomalies and trend analysis.
In the 3rd section Yu et al. 2022 presented the results as (1) correlation between field and satellite data, (2) spatial and seasonal distribution of different values of ice/firn/snow surface temperature on Greenland Ice Sheet and (3) anomalies in 20-years period, as well as, try to predict (4) future trend in the surface temperatures of the Greenland Ice Sheet.
As the 4th section entitled discussion, the Authors regard detecting melting periods on the Greenland Ice Sheet during studied period.
Finally, they drew a short conclusion about methodology and a summary about results.
Such a composition is quite logic and clear, but in my opinion there are some issues that should be discussed and commented by the Authors:
- A lack of Regional Setting description with some referencing to the regional factors affecting ice surface temperature. Additionally some basic description of GrIS is needed and its geological history/evolution that can be longer than 25 milllions of years (at least since Miocene!). In such a perspective the affirmation of GrIS melting made by authors melting looks like a joke or a complete lack of understand of this huge body of glacial ice.
- Ice surface temperature measurements / observation in the field is still very difficult task and more complicated than air temperature measurements, thus the results derived from the PROMICE Data Portal can be biased more than satellite data, while the Authors estimated the quality of the remote sensing on the basis of comparison to the PROMICE field data. My observations from Antarctic glaciers show that every item installed on glacier surface increases natural dynamics of glaciological process around the artificial body installed on glacier. To understand it the Authors should divide the AWSs into those installed below ELA (Equilibrium Line Altitude), thus melting into the glacier, and those above the ELA, thus inducing faster accumulation of snow. Such a differentiation enables critical view on the field data and possibly is a reason of widespread cold bias of MODIS IST over the GrIS, especially in the winter when the AWS are covered by snow in GrIS margin. Summit is always covered by snow, thus the AWS and MODIS reflects the snow temperature.
- The IST derived from MODIS with T above 0 Celsius grades are regarded as fake pixel values by Authors, thus they had removed. It is also visible on Fig.4c), where peaks of values above 0 look like cut intentionally by authors. Glacier surface can be covered by supraglacial water, supraglacial moraine deposits or ash blown from adjacent nunataks, thus it can be above 0°C.
- Cloud mask that limits MODIS data is another problem, because on open sky conditions you reach only specific weather type when snow/glacier surface receives solar radiation during the day, especially in the summer, or it experiences strong emissivity during the night, especially in the winter.
- Figure 1 should absolutely be reworked in order to better presentation of AWSs located in GrIS margin, ie. QAS, TAS, SCO, THU and NUK. Maybe some smaller locator maps are needed in these parts of GrIS. Replace “tundra” by “ice free areas” or “rocks and tundra”. Add meridians and latitude circles.
- Table 1. Develop full name of AWSs next to their acronyms and think about presentation the distance from GrIS margin and/or the elevation below/above the ELA, which is not mentioned anywhere in the MS, while it is a basic characteristisc of every glacier!
- Fig.3 can look more interesting when you organise it by distance from GrIS margin and/or the elevation below/above the ELA rather than SE, SW, NW, NE (and one AWS on Summit, surely hundreds m above ELA)
- I suggest linking the 2010 May anomaly (subsection 3.3) with eruption of Eyjafjallajökull in Iceland. A relevant weather anomaly occurred then in Europe as well and it happened shorty after the vulcanic eruption. I suggest reading of a paper: 10.1016/j.earscirev.2016.09.014
- I am afraid that searching for trends on the basis of 20 years period cannot be done honestly, because the period is too short
- The 2012 warm anomaly should be presented in more details, because this event was widely commented in the world, mainly by journalist who do not know what a glacier is and how it works. I remember that short-tem ablation occured then even on the summit of GrIS, what never happened in the inner part of Antarctic Ice Sheet. Maybe GrIS should be compared somehow to AntarIS. I suggest extracting from the MODIS a map presenting the max daily T values during 2012 warm anomaly.
I wish I could spend more time for reading some papers cited by the Authors, eg. by 2 x Fausto et al., because the topic looks interesting and the Authors can make a significant contribution to remote sensing and glaciology.
My detailed remarks and first impression from the MS reading are also available in enclosed PDF, however I have scanned the pages only where I wrote sth by red pen.
My overall decision is: major review.
I am looking forward to receive and review the improved version of the MS and possibly a future paper.
Comments for author File: Comments.pdf
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Thank you for including most of my remarks. I think your MS is worthy to be published right now
