the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Uncertainties originating from GCM downscaling and bias correction with application to the MIS-11c Greenland Ice Sheet
Abstract. The Marine Isotope Stage 11c (MIS-11c) interglacial is an enigmatic period characterized by a long duration of relatively weak insolation forcing, but is thought to have been coincident with a large global sea level rise of 6–13 m. The configuration of the Greenland Ice Sheet during the MIS-11c interglacial highstand is therefore of great interest. Given the limited data constraints, model-based analysis may be of use, but only if model uncertainties are adequately accounted for. A particularly under-addressed issue in coupled climate and ice sheet modeling is the coupling of surface air temperatures to the ice model. Many studies apply a uniform “lapse rate” accounting for the temperature differences at different altitudes over the ice surface, but this uniformity neglects both regional and seasonal differences in near-surface temperature changes. Herein we provide the first such analysis for MIS-11c Greenland that addresses these uncertainties by comparing 1-way coupled CESM and ice sheet model results from several different downscaling methodologies.
In our study, a spatially- and temporally-varying temperature downscaling method produced the greatest success rate in matching limited paleodata constraints, and suggests a peak ice volume loss from Greenland during MIS-11c of near 50 % compared to present day (~3.9 m contribution to sea level rise). This result is on the lower bound of existing data- and model-based studies, partly as a consequence of the applied one-way coupling methodology which neglects some feedbacks. Additional uncertainties are examined by comparing two different present-day regional climate analyses for bias correction of temperatures and precipitation, a spread of initialization states and times, and different spatial configurations of precipitation bias corrections. No other factor exhibited greater influence over the simulated Greenland ice sheet than the choice of temperature downscaling scheme.
-
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
(3193 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3193 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1367', Anonymous Referee #1, 28 Jul 2023
Crow and co-authors employed an offline ice sheet model to investigate the impact of spinup-time, climate forcing, and temperature downscaling on the evolution of the Greenland ice sheet during MIS11. The authors utilized CESM time-slice simulations as the basis for the climate forcing, which was further bias-corrected in various ways, and the impact thereof was investigated. The study highlights the substantial influence of both climate forcing bias correction (here RACMO vs MAR) and the approach used for temperature downscaling on the ice sheet evolution. Conversely, spinup-time and bias correction of precipitation were found to have minimal effects during MIS11. The authors also concluded that a spatially and temporally variable lapse rate for temperature downscaling is not only the most physically realistic approach but also yields the best results in reproducing paleodata constraints from Summit and DYE3. By using these two data constraints, the authors found that the Greenland ice sheet must have contributed at least 3.9 m sle at the peak of MIS11.
The manuscript is very well-written and nicely illustrated. The description of the model setups, coupling approach, and investigated sensitivities is mostly clear, but could benefit from some minor further additions. Furthermore, the compromises made with regard to coupling strategy, bias corrections, and ice sheet initialization are clearly explained, and it is convincingly demonstrated that these are reasonable given the scientific question the authors aim to answer. The study thoroughly evaluates the model's sensitivities to various sources of uncertainties, such as climate forcing and bias corrections, and extensively discusses the limitations of the employed models and approaches As such, I only have some minor comments, which I outline below.
Minor and technical points:
L80-85: You could also mention the recently published evidence from Camp Century being ice free during MIS11 (Christ et al., 2023, doi:10.1126/science.ade4248).
L117: Can you give more information which and how many time-slices you used.
L118: Please give a reference for the GHG concentrations used.
L125: Selected variables for what?
L259: Does the relatively coarse resolution also contribute to this (by not resolving small fjords etc.)?
L270: Please note relative to which dataset this biases exists. ERA or something else?
L276: Please elaborate a bit more how this scale factor was calculated. Is this the ratio between time-slice versus PD mean?
L338: Can you give the percentage of these 31 simulations that use the STV lapse rates?
L426: It seems that the difference between RACMO and MAR (Fig. 6) is similar impactful as the choice of lapse rate.
L436: Can you provide a figure with maps of the differences in temperature based on the different downscaling approaches for instances at 405 ka either in the main text or appendix/supplement? This would provide a better visualization of the impact of the different approaches.
Fig. 2: I think it could be more intuitive to depict blue colors as cold bias in CESM and red colors as warm bias, i.e., opposite to how it is shown now.
Tab. 1: Since either RACMO or MAR were used for both temperature and precipitation, it would make sense to combine both columns into one.
Citation: https://doi.org/10.5194/egusphere-2023-1367-RC1 -
AC1: 'Reply on RC1', Brian Crow, 31 Oct 2023
The authors would like to thank the anonymous reviewer for their thoughtful and positive response to our manuscript. We are pleased to see that the primary messages of our manuscript were well-received. Most suggestions will be fully implemented without further comment. The other specific points of constructive criticism have been addressed as follows:
- L80-85: Thank you for this reference, and we are glad to see that its results are consistent with ours. Mention will be added to the manuscript.
- L426: Comparing figures 5 and 6 illustrates that the magnitude of the difference in sea level contribution between the MAR and RACMO datasets varies widely depending on the lapse rate technique utilized. Therefore the choice of lapse rate technique is the first-order criteria affecting the melt magnitude and the dataset choice is secondary.
- L436: A figure of surface air temperatures for the four lapse rate types will now be included. We have chosen to show the 413 ka timestep, however, as this is during the height of the GrIS melting phase, and the contrasts are more clearly illustrated.
Citation: https://doi.org/10.5194/egusphere-2023-1367-AC1
-
AC1: 'Reply on RC1', Brian Crow, 31 Oct 2023
-
RC2: 'Comment on egusphere-2023-1367', Anonymous Referee #2, 06 Oct 2023
Review CP 2023-1367
Summary:
The study investigates the uncertainties from GCM downscaling and bias correction approaches during a highly interesting period of Marine Isotopic stage 11c (MIS11c). MIS11c is characterized by weak insolation which hint to a rather large global sea level rise of 6 to 13 m, i.e. the Greenland ice sheet during this period is of great interest. As proxy evidence is sparse during this period modelling approaches might give advice. Still, current modelling efforts suffer from uncertainties produced by (too) simple downscaling and bias correction approaches (e.g. constant lapse rate approach). In this study the author team try to tackle this issue by testing several different downscaling methods. The authors find that a spatially and temporally-varying method outperforms simpler methods showing that the ice volume loss over Greenland during MIS11c was roughly 50% compared to today. This is a contribution of 3.9 m of sea level equivalent and thus at the lower bound of the reconstructed changes.
General
The manuscript is well written and clearly structure. Overall, it tackles an important question on how we can improve our modelling strategies to assess Ice sheet loss in the past. Given the fact that the focus is on the he methods I see plenty of application possibles of this approach also for other interesting periods in the past and may be for future projections. Thus, the manuscript fits perfectly to Climate of the Past, dealing with a pressing issue. I recommend publication after some minor changes.
Comments
L19: Please change throughout the manuscript methodology to method. Note that “methodology” is the study of research methods.
Introduction: I think the authors could also give a very brief review on other strategies to deal with the problem, i.e., using dynamical downscaling to obtain better forcing data for the ice sheets. A recent study by a colleague shows the success:
Jouvet et al. 2023 Coupled climate-glacier modelling of the last glaciation in the Alps Journal of Glaciology, 1–15. https://doi.org/10.1017/jog.2023.74
L37: In the abstract the authors write 6 to 13 m, what is correct?
L183: The authors set the lapse rate to 7 K per km if the elevation differences are below 100 m. Clearly, given the coarse resolution the authors use this might be valid but there are hints that the lapse rate might have changed in past periods (see Jouvet et al. 2023). So, have you performed tests with other plausible lapse rate to assess the influence, e.g., 5 of 6 K per km.
L270: The bias correction is rather simple, maybe due to the coarse resolution it is the way to go but at least for precipitation the authors should discuss the potential short coming at the end of the manuscript. I could imagine that a height dependent precipitation bias correction could be a better choice, given that is it similar simple.
Citation: https://doi.org/10.5194/egusphere-2023-1367-RC2 -
AC2: 'Reply on RC2', Brian Crow, 31 Oct 2023
The authors would like to thank this anonymous reviewer for their very positive assessment of our manuscript and their recommendations for improvement. We are pleased to see that our core messages were well understood. The recommended changes are few and will therefore be addressed point-by-point below.
- L19: Respectfully, this is not correct. We follow long-established field precedent in using “methodology” to mean “a collection of methods or techniques used in a study or field of study,” a definition which is supported by various English dictionaries (e.g., Cambridge: METHODOLOGY | English meaning - Cambridge Dictionary).
- Introduction: Thank you to the reviewer for the recommended reference. A brief discussion of other coupling strategies, such as dynamical downscaling, will now be included.
- L37: The 6-13 m in the abstract refers to the overall MIS-11 sea level highstand. The reference on line 37 addresses the Greenland-only sea level contribution during MIS-11. This sentence will be lightly edited for clarity.
- L183: The fixed 7 K lapse rate for areas with small elevation deltas applies mostly to grid cells that are at least fractionally oceanic and therefore is of minimal consequence for surface temperatures of the ice sheet. Other lapse rates were therefore not tested.
- L270: The GSM already contains a physically-motivated orographic correction to precipitation inputs. The surface elevation and winds a few hundred meters above the surface are used to diagnose vertical velocities that lead fairly directly to the local orographic scaling factor with subsequent scaling to ensure regional mass-conservation. While this does not completely overcome the biases of the input data (i.e., CESM), a further orographic bias adjustment would be redundant. While the employed bias correction scheme is simplistic, the highly nonlinear nature of the dynamics responsible for precipitation do not lend themselves well to scalar adjustments. Discussion of this point has been expanded in section 2.6.
Citation: https://doi.org/10.5194/egusphere-2023-1367-AC2
-
AC2: 'Reply on RC2', Brian Crow, 31 Oct 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1367', Anonymous Referee #1, 28 Jul 2023
Crow and co-authors employed an offline ice sheet model to investigate the impact of spinup-time, climate forcing, and temperature downscaling on the evolution of the Greenland ice sheet during MIS11. The authors utilized CESM time-slice simulations as the basis for the climate forcing, which was further bias-corrected in various ways, and the impact thereof was investigated. The study highlights the substantial influence of both climate forcing bias correction (here RACMO vs MAR) and the approach used for temperature downscaling on the ice sheet evolution. Conversely, spinup-time and bias correction of precipitation were found to have minimal effects during MIS11. The authors also concluded that a spatially and temporally variable lapse rate for temperature downscaling is not only the most physically realistic approach but also yields the best results in reproducing paleodata constraints from Summit and DYE3. By using these two data constraints, the authors found that the Greenland ice sheet must have contributed at least 3.9 m sle at the peak of MIS11.
The manuscript is very well-written and nicely illustrated. The description of the model setups, coupling approach, and investigated sensitivities is mostly clear, but could benefit from some minor further additions. Furthermore, the compromises made with regard to coupling strategy, bias corrections, and ice sheet initialization are clearly explained, and it is convincingly demonstrated that these are reasonable given the scientific question the authors aim to answer. The study thoroughly evaluates the model's sensitivities to various sources of uncertainties, such as climate forcing and bias corrections, and extensively discusses the limitations of the employed models and approaches As such, I only have some minor comments, which I outline below.
Minor and technical points:
L80-85: You could also mention the recently published evidence from Camp Century being ice free during MIS11 (Christ et al., 2023, doi:10.1126/science.ade4248).
L117: Can you give more information which and how many time-slices you used.
L118: Please give a reference for the GHG concentrations used.
L125: Selected variables for what?
L259: Does the relatively coarse resolution also contribute to this (by not resolving small fjords etc.)?
L270: Please note relative to which dataset this biases exists. ERA or something else?
L276: Please elaborate a bit more how this scale factor was calculated. Is this the ratio between time-slice versus PD mean?
L338: Can you give the percentage of these 31 simulations that use the STV lapse rates?
L426: It seems that the difference between RACMO and MAR (Fig. 6) is similar impactful as the choice of lapse rate.
L436: Can you provide a figure with maps of the differences in temperature based on the different downscaling approaches for instances at 405 ka either in the main text or appendix/supplement? This would provide a better visualization of the impact of the different approaches.
Fig. 2: I think it could be more intuitive to depict blue colors as cold bias in CESM and red colors as warm bias, i.e., opposite to how it is shown now.
Tab. 1: Since either RACMO or MAR were used for both temperature and precipitation, it would make sense to combine both columns into one.
Citation: https://doi.org/10.5194/egusphere-2023-1367-RC1 -
AC1: 'Reply on RC1', Brian Crow, 31 Oct 2023
The authors would like to thank the anonymous reviewer for their thoughtful and positive response to our manuscript. We are pleased to see that the primary messages of our manuscript were well-received. Most suggestions will be fully implemented without further comment. The other specific points of constructive criticism have been addressed as follows:
- L80-85: Thank you for this reference, and we are glad to see that its results are consistent with ours. Mention will be added to the manuscript.
- L426: Comparing figures 5 and 6 illustrates that the magnitude of the difference in sea level contribution between the MAR and RACMO datasets varies widely depending on the lapse rate technique utilized. Therefore the choice of lapse rate technique is the first-order criteria affecting the melt magnitude and the dataset choice is secondary.
- L436: A figure of surface air temperatures for the four lapse rate types will now be included. We have chosen to show the 413 ka timestep, however, as this is during the height of the GrIS melting phase, and the contrasts are more clearly illustrated.
Citation: https://doi.org/10.5194/egusphere-2023-1367-AC1
-
AC1: 'Reply on RC1', Brian Crow, 31 Oct 2023
-
RC2: 'Comment on egusphere-2023-1367', Anonymous Referee #2, 06 Oct 2023
Review CP 2023-1367
Summary:
The study investigates the uncertainties from GCM downscaling and bias correction approaches during a highly interesting period of Marine Isotopic stage 11c (MIS11c). MIS11c is characterized by weak insolation which hint to a rather large global sea level rise of 6 to 13 m, i.e. the Greenland ice sheet during this period is of great interest. As proxy evidence is sparse during this period modelling approaches might give advice. Still, current modelling efforts suffer from uncertainties produced by (too) simple downscaling and bias correction approaches (e.g. constant lapse rate approach). In this study the author team try to tackle this issue by testing several different downscaling methods. The authors find that a spatially and temporally-varying method outperforms simpler methods showing that the ice volume loss over Greenland during MIS11c was roughly 50% compared to today. This is a contribution of 3.9 m of sea level equivalent and thus at the lower bound of the reconstructed changes.
General
The manuscript is well written and clearly structure. Overall, it tackles an important question on how we can improve our modelling strategies to assess Ice sheet loss in the past. Given the fact that the focus is on the he methods I see plenty of application possibles of this approach also for other interesting periods in the past and may be for future projections. Thus, the manuscript fits perfectly to Climate of the Past, dealing with a pressing issue. I recommend publication after some minor changes.
Comments
L19: Please change throughout the manuscript methodology to method. Note that “methodology” is the study of research methods.
Introduction: I think the authors could also give a very brief review on other strategies to deal with the problem, i.e., using dynamical downscaling to obtain better forcing data for the ice sheets. A recent study by a colleague shows the success:
Jouvet et al. 2023 Coupled climate-glacier modelling of the last glaciation in the Alps Journal of Glaciology, 1–15. https://doi.org/10.1017/jog.2023.74
L37: In the abstract the authors write 6 to 13 m, what is correct?
L183: The authors set the lapse rate to 7 K per km if the elevation differences are below 100 m. Clearly, given the coarse resolution the authors use this might be valid but there are hints that the lapse rate might have changed in past periods (see Jouvet et al. 2023). So, have you performed tests with other plausible lapse rate to assess the influence, e.g., 5 of 6 K per km.
L270: The bias correction is rather simple, maybe due to the coarse resolution it is the way to go but at least for precipitation the authors should discuss the potential short coming at the end of the manuscript. I could imagine that a height dependent precipitation bias correction could be a better choice, given that is it similar simple.
Citation: https://doi.org/10.5194/egusphere-2023-1367-RC2 -
AC2: 'Reply on RC2', Brian Crow, 31 Oct 2023
The authors would like to thank this anonymous reviewer for their very positive assessment of our manuscript and their recommendations for improvement. We are pleased to see that our core messages were well understood. The recommended changes are few and will therefore be addressed point-by-point below.
- L19: Respectfully, this is not correct. We follow long-established field precedent in using “methodology” to mean “a collection of methods or techniques used in a study or field of study,” a definition which is supported by various English dictionaries (e.g., Cambridge: METHODOLOGY | English meaning - Cambridge Dictionary).
- Introduction: Thank you to the reviewer for the recommended reference. A brief discussion of other coupling strategies, such as dynamical downscaling, will now be included.
- L37: The 6-13 m in the abstract refers to the overall MIS-11 sea level highstand. The reference on line 37 addresses the Greenland-only sea level contribution during MIS-11. This sentence will be lightly edited for clarity.
- L183: The fixed 7 K lapse rate for areas with small elevation deltas applies mostly to grid cells that are at least fractionally oceanic and therefore is of minimal consequence for surface temperatures of the ice sheet. Other lapse rates were therefore not tested.
- L270: The GSM already contains a physically-motivated orographic correction to precipitation inputs. The surface elevation and winds a few hundred meters above the surface are used to diagnose vertical velocities that lead fairly directly to the local orographic scaling factor with subsequent scaling to ensure regional mass-conservation. While this does not completely overcome the biases of the input data (i.e., CESM), a further orographic bias adjustment would be redundant. While the employed bias correction scheme is simplistic, the highly nonlinear nature of the dynamics responsible for precipitation do not lend themselves well to scalar adjustments. Discussion of this point has been expanded in section 2.6.
Citation: https://doi.org/10.5194/egusphere-2023-1367-AC2
-
AC2: 'Reply on RC2', Brian Crow, 31 Oct 2023
Peer review completion
Journal article(s) based on this preprint
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
280 | 95 | 27 | 402 | 19 | 16 |
- HTML: 280
- PDF: 95
- XML: 27
- Total: 402
- BibTeX: 19
- EndNote: 16
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
Lev Tarasov
Michael Schulz
Matthias Prange
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3193 KB) - Metadata XML