This paper provides a short communication presenting velocity and surface elevation changes over Thwaites glacier in Antarctica. Despite the presented data range between 2012-2020, new results are related to the period 2017-2020 only. The main finding of the paper is related to the underwater cavity found in previous studies at the main trunk of Thwaites glacier. The authors claim that melt at the cavity stopped during the period 2017-2020. However, these findings are only based on InSAR surface elevation changes which is typically 10 times less accurate than altimeter based surface elevation measurements.  DEM only based results can be misleading when adopted for calculating melt rates and grounding line retreats without supporting data such as DInSAR or RADAR sounders.

The authors also do not explain the methodology used for calculating melt-rates (as an example in figure 2 ice bottom crosses the bedrock). This lack makes it hard to evaluate the results. Some statements seem to be overly speculative (see comments below) and are not proved or confirmed using independent and complementary measurements.

These issues reduce my overall confidence in the main results. The presentation is also unclear at times. I suggest major revisions are required for readers to have confidence in the reliability of results and the conclusions that are presented.

Line 45 – “We tied the adjacent DEMs to their cotemporal neighbours using a point within the scene overlap. By choosing this point over a relatively flat region, height errors resulting from geolocation errors were minimised”

It is not clear to me why the authors need to coregister adjacent DEMs if they use IceSAT for calibrating the single tile. Moreover this approach, if not performed correctly might lead to errors related to geocoding or mi-registration (see schematics in the PDF version). Hence a single tie point seems to be a weak approach for tackling this problem.

54-55 “Small tidal Ranges” of 0.5 meters would correspond to abut 5 meters of difference when considering multiplying by the flotation factor. This factor should be taken into account if you are looking at grounding zones or a more solid justification of why this was not taken into account should be provided.

65 Why do you need tie points described on line 45 if you can calibrate data with icesat?

69 TerraSAR-X SLCs (2012–2014) why only this period and not the entire dataset? Moreover, you can do pixel tracking with DEMs itself if pairs far apart in time are characterized by Low SNR levels in the Pixel Offset maps.

73 What does sampling every 100 m means? Do you pick a cross-correlation window every 100 m (for Sentinel) and 40 m for (TSX)? What are  other parameters like Window size and widow search used for different sensors? Depending on the used parameters the velocity maps will have different accuracies

79 It does not look like there is good agreement betweem the DEM based results compared to the GL west the grounding line (Fig 1.a). The TDX backscatter intensity covers the elevation of floating areas. How the DEM derived grounding lines have been obtained has not been explained anywhere in the manuscript.  This part requires an extensive rewriting.

85-89 This statement is not supported by the data since the DEM based Grounding lines are 2 orders of magnitude less sensitive compared to the InSAR based Grounding lines. This also is shown in Figure 1a West of thwaites main trunk where no grounding line is detected by the “DEM based” grounding line.  As explained in Milillo et al 2019 a grounding line retreating could show up as a surface uplift simply because bending forces at the grounding line get released once the grounding line retreats. This effect would alter height above flotation calculations together with the fact that the grounding line is not in hydrostatic equilibrium. Since you are not comparing Grounding lines obtained with the same measurements this statement seems too speculative.

90-92 This sentence is not supported by any data in the manuscript. How do you identify the cavities? This seems to be part of the discussion and not of the results.

109 how do you calculate the cavity volume? This is not explained in the manuscript

129-130 This sentence is not supported by observations or data and seems overly speculative.

Figure 1a Instead of surface elevation please show height above flotation since you are comparing grounding line measurements. Add legend with Grounding line acquisition dates

Figure 2 Despite the nice colorbar It seems counterintuitive to look at the bottom of the ice extending below the bedrock level. Since no methodology for retrieving these thickness change measurements has been described it is really hard to evaluate this figure and comment on it.

The presented brief communication article deals with the formation and temporal evolution of a previously identified cavity close to the grounding line at Thwaites glacier. Recent studies identified tipping points for a continued grounding line retreat for Pine Island glacier also located at WAIS [1]. Similarly Thwaites glacier has also been implicated to experience continued retreat of the grounding line [2]. Due to the fast ice flow at the main trunk of Thwaites it is currently challenging to update grounding line positions from InSAR acquisitions. It only remains possible with a small temporal baseline (COSMOS-Skymed, Milillo 2019). Therefore, a detailed time series of grounding line positions is of high interest to the scientific community as it allows to under investigate the melt processes and timescales of grounding line migration directly at the grounding lie, although a derivation of the grounding line from height above floatation is less accurate than InSAR derived grounding lines. It has to be noted that accurate bathymetry and density assumptions are crucial for correct grounding line positions. The used TDM data (if properly calibrated) is an accurate enough reference for the surface elevation.

From the time series of height above floatation measurements the authors concluded that the previously reported cavity remained stable in height and extent and therefore the grounding line position also remained at a position of a slightly upward sloping bed which is predicted by coupled ice–ocean models.

I have some points that need further addressing before publication

- My first comment concerns the vertical calibration of the TDM time series. It is currently only explained in a few sentences. It would be preferable to use more than one IceSAT-2 measurement for calibrating the TDM scene to the IceSAT data. I suggest adding the used IceSAT-2 track in one of the Figures for a better overview. If surface elevations over the crevassed floating parts are used, it is important to calibrate the TDM data also in areas of limited or no signal penetration. A statement about the surface roughness, or distance to the area of investigation should be included. If the surface is rough and crevassed I would see no problem in selecting the area as it was done in the article but the argument is missing. Also a statement about the size of the calibration area is missing. What diameter does the footprint of  the IceSat measurement have in the ATL06 data and how what size of TDM area was it compared against? Regarding the calibration of neighboring scenes in the range direction one hast to be careful to also include the TDM baseline uncertainties in the error budget of the adjacent scenes, as the two scenes are not from the same track and can be characterized by different baseline errors. A baseline uncertainty of 1mm depending on the height of ambiguity adds elevation uncertainties in the order of 1m [3]. Depending on the used method for vertical calibration to IceSAT-2 a tilt in range could be remaining and propagate to the neighboring scene.

- The actual derived grounding line from height above floatation is not displayed in Figure 1. The caption states  only MEaSUREs (purple: 1996 and yellow: 2011) and Milillo et al. (white: 2019). A time-series of 2D grounding lines would strengthen the argument of the suitability of height above floatation in this case, especially as it allows for a comparison with InSAR derived grounding lines over the whole area. L. 79 suggests that this was done. If a 2D representation of the grounding line time-series does not reproduce previous InSAR results over the entire area, it has to be stated that the analysis is restricted to the area of the cavity. In this case, results from height above floatation could be calibrated to the InSAR grounding line position.

- The discussion and especially the link to coupled models L 115-120 is difficult to understand. For me the physical process of why a stable grounding and cavity volume is reached after several years (how many?) is not entirely clear. Is this predicted by these models because they take ocean circulation of warm water in the cavities into account? If other models are used, would they predict a growing cavity and subsequent grounding line retreat (L. 120)? How the increasing velocity Fig 3, A1 are used in the arguments from L. 124-134 is not clear. I do not understand the meaning of this sentence "However, bed topography and ice-thickness close to floatation can superimpose rapid local change on the background long-term evolution of Thwaites and other WAIS glaciers in ASE"

Overall the article is of high scientific interest and well presented with clear language. The raised concerns require major revisions. 

[1] Rosier, Sebastian H. R., Ronja Reese, Jonathan F. Donges, Jan De Rydt, G. Hilmar Gudmundsson, und Ricarda Winkelmann. 2021. „The Tipping Points and Early Warning Indicators for Pine Island Glacier, West Antarctica“. _The Cryosphere_ 15 (3): 1501–16. [https://doi.org/10/gjnwhg](https://doi.org/10/gjnwhg).

[2] Joughin, I., Smith, B. E., and Medley, B.: Marine ice sheet collapse potentially under way for the Thwaites Glacier basin, West Antarctica, Science, 344, 735–738, [https://doi.org/10.1126/science.1249055](https://doi.org/10.1126/science.1249055), 2014.

[3] Rizzoli, Paola, Michele Martone, Carolina Gonzalez, Christopher Wecklich, Daniela Borla Tridon, Benjamin Bräutigam, Markus Bachmann, u. a. 2017. „Generation and Performance Assessment of the Global TanDEM-X Digital Elevation Model“. _ISPRS Journal of Photogrammetry and Remote Sensing_ 132 (Oktober): 119–39. [https://doi.org/10.1016/j.isprsjprs.2017.08.008](https://doi.org/10.1016/j.isprsjprs.2017.08.008).

Line comments:

- L. 4 continued

- L. 19 mention the used data ERS, COSMO-Skymed for deriving the grounding lines

- L. 44-46 (See general comments)

    - Choosing only one tie point is not robust

    - Combining adjacent across trade scenes in the overlap region includes baseline errors of in the order of 1m

    - Depending on the surface properties of chosen point there might be an elevation bias due to signal penetration.

    - Is the same point IceSAT-2 measurement used for the entire time series. If so, thinning rates should be close to 0 and quantified at this location from an independent source.

    - Show IceSat 2 track on Fig 1

- L. 54 What is the result of 0.5m tidal variation in thickness change?

- L. 68. Combining errors: baseline, tidal range, TDM orbit. height above floatation from the two scenes will be characterized by different errors

- L. 78 Could you calibrate f_a on the previous InSAR grounding line positions

- L. 80 loss of 50 to 60m

- L. 88 Quantify value. How many meters above flotation is reported. Could this be explained by erroneous bathymetry?

- L. 97 good agreement to InSAR grounding line locations

- L. 112 thick → deep - what does imprinted with bed topography mean?

- L. 130 Not justified - Discussion is hard to follow

- Fig. 1 

    - Show IceSAT-2 tracks

    - Missing 2D time series of height above floatation derived grounding line positions

    - A legend would be helpful. MEaSUREs (purple, yellow), Milillo et al. (white)

- Fig2: 

    - Reword caption: Surface elevation and basal elevation inferred from hydrostatic thickness. The thickness itself is not plotted.

    - Can you also quantify the scaling factor as it was used in the study here. 

    - I cannot distinguish the colors of the arrows. The arrows should be labelled.