the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 Oct 2025
Bottom mixed layer derivation and spatial variability over the central and eastern abyssal Pacific Ocean
Abstract. The bottom mixed layer (BML) of the abyssal ocean regulates heat exchange between the deep interior and seafloor, driving water-mass transformation and influencing global circulation. Spatial variability of the BML was examined in the under-sampled abyssal Pacific Ocean using surface-to-seafloor temperature and pressure observations over 4 months in 2023–24. Given the typical decadal repeat rate of global hydrographic sections, subdecadal variability in the abyssal ocean has remained poorly resolved. While constrained in coverage, our observations contribute towards filling this gap for the central and eastern abyssal Pacific Ocean. Four methods were used to determine the BML thickness, with the threshold method providing the most reliable estimates. The mean BML thickness was 226 ± 172 with added repeat hydrographic sections providing context and additional data points. At each BML data point we determined the slope, the terrain roughness and the extracted predicted internal tide energy dissipation (over five different low-mode processes and high-mode local processes) at 50 km scales from publicly available datasets. These factors were input into a Random Forest Regressor (RF) model, the first time machine learning techniques have been applied to investigate BML thickness. The RF feature importance scores identified bottom depth, total internal tide energy dissipation, followed by slope, as the strongest predictors of BML thickness, revealing the importance of low-mode internal wave energy losses in this abyssal setting. Targeted and sustained observations near the seafloor at gateway regions of abyssal pathways are vital for understanding energy exchange that influences meridional overturning circulation. Our results highlight a regime where sustained low-mode internal tide energy loss, modulated by topographic slope and depth, governs the BML thickness in the abyssal Pacific. However, the rate at which BML thickness changes over time and the processes that cause these changes remain key unresolved factors.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(19464 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(19464 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4709', Anonymous Referee #1, 03 Nov 2025
- AC2: 'Reply on RC1', Jessica Kolbusz, 04 Dec 2025
-
RC2: 'Comment on egusphere-2025-4709', Anonymous Referee #2, 19 Nov 2025
Review of “Bottom mixed layer derivation and spatial variability over the
central and eastern abyssal Pacific Ocean”
The authors use a series of historical and novel abyssal observations in the Eastern Pacific and report on the spatial variability of the bottom boundary layer. This is achieved through testing a series of methods to identify the bottom boundary layer. Following this they use a machine learning technique to identify the contributions to that variability from the bottom topography (depth, slope and roughness) and internal tide dissipation. To the best of my knowledge, this manuscript would be the first publication applying this machine learning approach to the bottom boundary layer. I believe that, following some additional work to strengthen the paper as described below, it would be worthy of publication in Ocean Science.
It is not clear to me why the authors made the choice to infer the salinity using the GMM. If I understood line 112 correctly there is a CTD on the bottom lander system so I expect they have measured salinity? If there is something I missed here, I think it should be made clearer in the text. At the very least, I would like to see some of derived salinity profiles or T-S diagrams with a comparison with what salinity data is available (either from the cruises or from climatology) to give confidence.
It is useful somewhere to identify that there is potentially an important difference between a mixed layer defined by hydrography and an active mixing layer where turbulence is acting over the layer. This is very important when inferring the importance of the boundary layer in e.g. water mass transformation. The surface mixed layer community have started to address this nuance and we should too.
The machine learning technique (as I understand it) is simply identifying where the bottom mixed layer height has the same spatial structure as the potential contributions. This leaves me a little concerned about how to interpret the results given the apparent overlap between some of the inputs. For example, the two main components identified (the bottom depth and the total dissipation) have very similar distributions in Figure 7. I think that the uncertainties and limitations of this method need to be discussed more in the text and acknowledged in the interpretation of the results.
A thought that you might want to bring into your discussion (optional), if the machine learning is identifying the total dissipation as an important control (subject to my concerns above) does that imply that the hydrographic definition of the boundary layer is likely similar to the active mixing layer for the bottom boundary layer for this region. It also implies strong local control of the hydrographic boundary layer thickness rather than control by relatively far-field topography (e.g. set by the sill depth between basins). This could be an important result.
Line 129 – 130 > It is not clear to me what this sentence means
Line 133 – 136 > These sentences seem repetitive
Line 145 > Define what the difference is relative to (seafloor / deepest part of profile / etc)
Line 195 > Remove remained
Line 222 > repetition of datasets
LIne 301 > I didn’t understand why the importance of dissipation was enhanced here. If I understood correctly both the dissipation and bottom depth are acting in the same direction to reduce the BML so either could be driving the change?
Citation: https://doi.org/10.5194/egusphere-2025-4709-RC2 - AC1: 'Reply on RC2', Jessica Kolbusz, 04 Dec 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4709', Anonymous Referee #1, 03 Nov 2025
In this manuscript (MS), the authors utilize novel field data to reveal the characteristics of bottom mixed layer (BML) thickness in the central and eastern Pacific and identify key controls on BML variability (ocean depth, total internal tide dissipation, slope). The RF analysis in the MS is first application of machine learning to BML thickness, identifies physically intuitive predictors, which is a notable strength. Several issues related to the RF regression and result interpretation need refinement to enhance the manuscript’s scientific rigor and impact. The detailed comments are provided below.
- The authors use GMM to predict practical salinity for TPT profiles using nearby GO-SHIP data, but no quantitative comparison is provided between predicted salinity and independent measurements (e.g., discrete water samples from TPT, or collocated GO-SHIP profiles not used for model training).
- In the MS, the authors tried to characterize and explain the BML patterns in the central and eastern Pacific that covers the fracture zones, but the data used are mainly located in the central Pacific. In the RF regression, the input GO-SHIP data are selected within 10° of the TPT measurements, which could create a data imbalance that risks biasing the model toward GO-SHIP’s spatial characteristics.
- In the RF regression, stratification and shear (or the state of stability) are important factors for determining the BML thickness, but are not considered as potential influencing parameters.
- The relationship between BML thickness and depth, total dissipation, slope, etc., may be more intuitively displayed using scatter plots or similar methods.
- Line 119, the authors mentioned Appendix A1, however in the Appendix A, it is about GMM, the information for TPT profiles can not be found.
- The MS reports key spatial patterns but provides limited physical explanation for these differences. For example, the authors note cooler, saltier AABW near Hawaii vs. fresher NPDW at the equator, but do not explicitly connect this to stratification (a key control on BML mixing). Stronger stratification in AABW regions should suppress mixing and thin the BML—consistent with thinner BML south of Hawaii.
- Figure 2: the locations of each profiles are not shown, making it difficult to link BML thickness to regional features. The “visual interpretation” line in Figure 2c is helpful but should be standardized across all subplots for consistency. The x-axis ranges of some subplots are too large to visually identify the BML thickness. Also the solid dots showing BML results of different methods could cover some critical features of the profiles.
- Equation 1: The variable h1 is not defined. It likely represents the seafloor depth, but this should be explicitly stated.
- The first mention of “σ₄” (Section 2.3) does not explain it is potential density referenced to 4000 dbar.
- Line 146: The citation of Hogg et al. (19821) contains a typo.
- Line 149 and 167: Appendix A2 should be Appendix B.
- Line 159: The authors refer to “Douglas-Peuchker” is a typo.
- In appendix B, the authors justify the 0.003°C threshold for BML derivation using “highest mean QI” but do not show some example profiles with BML results regarding different thresholds. QI might be high when the mixed layer results are shallower than the actual ones.
- In Figure A2, there are no subplots (c-e).
Citation: https://doi.org/10.5194/egusphere-2025-4709-RC1 - AC2: 'Reply on RC1', Jessica Kolbusz, 04 Dec 2025
-
RC2: 'Comment on egusphere-2025-4709', Anonymous Referee #2, 19 Nov 2025
Review of “Bottom mixed layer derivation and spatial variability over the
central and eastern abyssal Pacific Ocean”
The authors use a series of historical and novel abyssal observations in the Eastern Pacific and report on the spatial variability of the bottom boundary layer. This is achieved through testing a series of methods to identify the bottom boundary layer. Following this they use a machine learning technique to identify the contributions to that variability from the bottom topography (depth, slope and roughness) and internal tide dissipation. To the best of my knowledge, this manuscript would be the first publication applying this machine learning approach to the bottom boundary layer. I believe that, following some additional work to strengthen the paper as described below, it would be worthy of publication in Ocean Science.
It is not clear to me why the authors made the choice to infer the salinity using the GMM. If I understood line 112 correctly there is a CTD on the bottom lander system so I expect they have measured salinity? If there is something I missed here, I think it should be made clearer in the text. At the very least, I would like to see some of derived salinity profiles or T-S diagrams with a comparison with what salinity data is available (either from the cruises or from climatology) to give confidence.
It is useful somewhere to identify that there is potentially an important difference between a mixed layer defined by hydrography and an active mixing layer where turbulence is acting over the layer. This is very important when inferring the importance of the boundary layer in e.g. water mass transformation. The surface mixed layer community have started to address this nuance and we should too.
The machine learning technique (as I understand it) is simply identifying where the bottom mixed layer height has the same spatial structure as the potential contributions. This leaves me a little concerned about how to interpret the results given the apparent overlap between some of the inputs. For example, the two main components identified (the bottom depth and the total dissipation) have very similar distributions in Figure 7. I think that the uncertainties and limitations of this method need to be discussed more in the text and acknowledged in the interpretation of the results.
A thought that you might want to bring into your discussion (optional), if the machine learning is identifying the total dissipation as an important control (subject to my concerns above) does that imply that the hydrographic definition of the boundary layer is likely similar to the active mixing layer for the bottom boundary layer for this region. It also implies strong local control of the hydrographic boundary layer thickness rather than control by relatively far-field topography (e.g. set by the sill depth between basins). This could be an important result.
Line 129 – 130 > It is not clear to me what this sentence means
Line 133 – 136 > These sentences seem repetitive
Line 145 > Define what the difference is relative to (seafloor / deepest part of profile / etc)
Line 195 > Remove remained
Line 222 > repetition of datasets
LIne 301 > I didn’t understand why the importance of dissipation was enhanced here. If I understood correctly both the dissipation and bottom depth are acting in the same direction to reduce the BML so either could be driving the change?
Citation: https://doi.org/10.5194/egusphere-2025-4709-RC2 - AC1: 'Reply on RC2', Jessica Kolbusz, 04 Dec 2025
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 239 | 87 | 26 | 352 | 16 | 16 |
- HTML: 239
- PDF: 87
- XML: 26
- Total: 352
- BibTeX: 16
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Devin Harrison
Nicole Jones
Joanne O'Callaghan
Taimoor Sohail
Todd Bond
Heather Stewart
Alan Jamieson
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(19464 KB) - Metadata XML
In this manuscript (MS), the authors utilize novel field data to reveal the characteristics of bottom mixed layer (BML) thickness in the central and eastern Pacific and identify key controls on BML variability (ocean depth, total internal tide dissipation, slope). The RF analysis in the MS is first application of machine learning to BML thickness, identifies physically intuitive predictors, which is a notable strength. Several issues related to the RF regression and result interpretation need refinement to enhance the manuscript’s scientific rigor and impact. The detailed comments are provided below.