General comments:

This study uses a combination of historical CTD profiles (1980-2023) and CTD and velocity data from two moorings (2010-2022) to investigate long-term changes in the deep (~2500m depth) water masses in Fram Strait and its surroundings (the Greenland Sea and the Eurasian Basin of the Arctic Ocean). The deep Arctic Ocean is sparsely sampled and Fram Strait is the only gateway for exchange of deep water between the Arctic Ocean and the rest of the World Ocean. Hence, this study provides an important contribution to understanding how the deep Arctic Ocean has evolved and may change in the future. In particular, the results show a strong warming of the deep water in the Greenland Sea, which since ~2018 has become warmer than the deep water in the Arctic Ocean and may provide as a deep heat source. The manuscript is well structured and written, has clear figures, with appropriate analyses that support the conclusions. I only have some minor comments/questions that should be addressed before the manuscript is published.

Specific comments and technical corrections:

L3: Define “deep ocean”

L5: Suggest replacing “upstream basins” with “surrounding basins” or similar, as the basins can also be “downstream” in terms of the flow.

L22: Rapid warming of GSDW or less GSDW in the mixture? Related to my comment on the water mass definition (L76-77) below.

L31: Do you mean mixing in Fram Strait or exchange across the strait? I think exchange is a better word here.

L36: “… GSDW now being a lighter …”: I’m missing a reference here. Is this based on Somavilla et al., 2013? Then the word “now” is not appropriate. Rather specify when the transition happened.

L51: “knowledge of …, since 2012/2014, is lacking.”: why do you include two different years?   

L53-61: I found the start of this paragraph (L53-57) a bit difficult to follow because the main goal and focus of the paper are not mentioned in the introduction yet. It would be good to clarify this in the beginning of this paragraph (instead of later in L58-59).

L73: Is there a typo in the east-west extension of the GS box in the text? It says 12W-12E (same as the FS box), but on the map it doesn’t look like the GS box goes to 12E (rather 0E, which makes more sense too, in terms of excluding the NS).

L76: exact choice “of” depth levels

L76-77: I agree that it is suitable to use a property-independent definition to look at deep water changes in specific regions. However, I think it would be good to clarify that the deep waters you look at contain (are mixtures of) water masses with different origin. GSDW is often referred to in the literature as a class of water that was formed in the Greenland Sea prior to the 1980s. Your GSDW definition likely contains an increasing amount of other water masses such as EBDW and GSAIW. You do discuss this later on in the discussion but it would be good to include one or two sentences here to avoid confusion.  

L80-83: Please be consistent with the number of decimals you present for the salinity values (and temperature values). This is also the case in other parts of the manuscript (e.g., section 2.5 and the first paragraph of section 3.1).

L81: … which “corresponds to a” change of … in Practical Salinity …    (same for temp.)

L95-98: Does this mean that you exclude deployments during the 2010-2022 period or are the deployments with Aanderaa RCMs before 2010 (which is why you start in 2010)? This is not clear. Also, the Aanderaa RCMs are dataloggers/platforms with point current meters and the option to install different types/versions of temperature sensors (etc.) with different specs. Which temperature sensor (product number) has the accuracy you refer to of +/-0.05degC?

Table 1: “Not all data is always available during the deployments.” I think this needs some elaboration. Is it because of gaps in the deployments, sensor failures, or because you excluded deployments with RCMs or other reasons?

L121: “… compare our results to previous studies, …”: which results? I’m missing one introductory sentence about why you are doing the cross-correlation analysis (like the first sentence you have in section 2.4, L139-140).

L124: Remove one “the” in “We define the the upper and lower bounds…”.

L124-126: “We define the upper and lower bounds by …, to get estimates of the upper and lower bounds of water mass properties.”: The last part of this sentence is not needed/is a repeat of the start (ie. remove the part after “to get estimates ….”).

L152-153: “The results were mostly robust to the choice of year (not shown), ..”: It is difficult to know what “mostly robust” means. Could you be more quantitative or give an example?

L157: …, which have “an” initial accuracy …

L175-176 and Fig 2: What does the density change correspond to in terms of potential density anomaly referenced to the surface? (particularly for the DW in the Greenland Sea).

L176-177: It is not clear from the data in fig.2d that the GSDW temp is higher than EBDW after ~2015. That is, I would just state that they have converged here. Evidence of larger GSDW temp comes primarily from the mooring data in Fram Strait and the cross-correlation analyses later.

L217: either “flow switches” or “flows switch”

Fig. 4 caption: … as defined “by the regime shift analysis” in Sec. 2.4.

L226: Remove “in fact” before “EBDW-dominated”.

L230-231: If I understand it correctly it is primarily the seasonal cycle in velocity combined with the different rates of warming of GSDW and EBDW that leads to the stronger seasonality in temperature. What about seasonality in temperature in the upstream water masses?

Equation 2: multiply by 1/2

L242: … direction of the “deep” flow in Fram Strait …

L248: … associated “with” sampling …

L282: “However, the possibility remains that GSAIW is replacing GSDW.” Could you plot typical properties of GSAIW to check this? What about geothermal heating? Could this be another source of the increased temperature at 2500m?

L312: … play for understanding the changing Arctic. Remove “our” and “on” before and after “understanding”, respectively.