the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Simulation of a former ice field with PISM – Snežnik study case
Abstract. In this paper we present a reconstruction of climate conditions during the Last Glacial Maximum on a karst plateau in Dinaric Mountains (southern Slovenia) that bares evidence of glaciation. The reconstruction merges geomorphological ice limits, classified as either clear or unclear, and computer modelling approach based on Parallel Ice Sheet Model, which is an established numerical model for simulating glacier dynamics ranging from ice sheets to alpine glaciers. Based on extensive numerical experiments, where we studied the agreements between simulated and geomorphological ice extent, we propose to use a combination of high resolution precipitation model that accounts for orographic precipitation combined with simple elevation based temperature model. The geomorphological ice extent can be simulated with climate around 6 °C colder than modern and with a lower than modern amount of precipitation, which matches other state-of-the art climate reconstructions for the era. The results indicate that orographic precipitation model is essential for accurate simulation of the Snežnik with moist southern winds from the nearby Adriatic Sea having predominant effect on the precipitation patterns. Finally, this study shows that transforming climate conditions towards wetter and warmer or drier and colder does not significantly change conditions for glacier formation.
- Preprint
(5622 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-544', Anonymous Referee #1, 05 Apr 2024
This manuscript presents simulations of a small plateau glacier in present-day Slovenia during the last glacial maximum (LGM). This work is motivated by earlier geomorphological reconstructions of past glaciation. It aims to answer the question whether these reconstructions can be used to estimate the regional climatic conditions during the LGM.
While the study mostly fails to provide a robust reconstruction of glacial climate, it succeeds in establishing a methodology that may be used for similar attempts in the future. Despite the large uncertainties inherent to paleoclimate studies, the manuscript describes a solid methodology and and a mostly complete discussion of its shortcomings. The individual steps of the modeling approach and the evaluation are described in great detail and the text follows a clear logical structure. The figures support the text optimally. I greatly enjoyed reading this well-written piece of work.
I have three major remarks:
1) I think the main contribution of this study is not the actual temperature reconstruction, because it is rather imprecise and lacks an uncertainty estimate. Instead, I concur with the authors that "This research has successfully established a [..] framework for the assessment of palaeoglacial simulations that integrates [..] geomorphological deduced ice boundaries to improve the accuracy of model results" (l471ff). I think the discussion of results and the conclusions should reflect this change of emphasis and provide a deeper discussion of the shortcomings as well as how they can be ammended. A few open questions include the choice of surface mass balance model (PDD) and in particular how the PDD factors were chosen. Are they left at their default values or were they adjusted to this particular domain? How exactly is the temperature reconstruction impacted by the incomplete knowledge of precipitation? Why does the model consistently fail to simulate certain parts of the reconstructed glacier area?2) Virtually all simulations include a substantial glaciation of the mountain to the southeast of Snežnik, which I believe is called Ceclje. This is not too surprising as the elevation of the surrounding terrain is similar and so are the climatic conditions. However, the geomorphological reconstructions show a strict boundary and thus a "forbidden area" that penalizes the simulations for having ice there. How certain are you that Ceclje was not glaciated during the LGM? Should the possibility of its glaciation not be reflected in the skill metric? And, more generally, the manuscript takes the approach of using the geomorphological data to inform the modeling, but the opposite could also be done. I think the possibility of a glaciation to the southeast should at least be discussed.
3) Lastly, I would like to see a discussion of how the reconstructions could be improved. Ideally, geomorphologists and other field-going scientists could used the simulations results to inform their work and thereby help to better constrain the next iteration of the reconstruction. What kind of data and from where would be most valuable?
Minor remarks:
figures: I would prefer the original model grid to be visible in all figures and therefore recommend not using interpolation in figures 4 and 5. I like the pixelated nature of figure 6 etc. better.line 3: There is no need to justify the use of PISM. I would remove ", which is an established...".
line 9: "Snežnik" is not yet introduced and it is unclear what it means at this point.
line 46: I think the introduction should include one paragraph with a preview of the main results.
line 65: "ARSO" has not yet been introduced at this point in the text.
tables 1-3: I think they can be combined into one.
figure 1: Why is the domain topography shown at a resolution of 25 m if the simulations use at most 50 m? Again, I would like to see the data on the original model grid without interpolation.
figure 2: The two shades of green are not optimal.
line 360ff: I am not sure this finding is very unexpected or important. Changes in temperature and precipitation may balance out and the effects are somewhat linear for small changes. This is how a Taylor expansion works.
figure 9: If I understood correctly, the red curve represents a simulation that stays at 400 m resolution even if the horizontal axis suggests otherwise. This is not very clear from either the legend of the figure caption and should be improved.
line 394: Is this result the same for all precipitation models? Which one was used here?
line 424: I think this relationship between T and P makes the original goal of reconstructing climate impossible without additional data that constraints either T or P. Most problems in the geosciences are underdetermined, but this one to a degree that needs to be addressed explicitly.
line 475: I think calling the simulations "consistent with emperical geomorphological reconstructions" is a bit too overconfident. There are systematic biases.
line 484f: It should be noted that this relative insensitivity is found for the particular case at hand, not in general.
line 487: "The study reveals that..." This statement should be removed. This fact is well known on a general level and the SMB model is not detailed enough to provide a deeper insight into the T/P compensation effect.
line 493f: "This model is consistent with geomorphological field data..." Again, I think the simulation results are not good enough for that. The methodology allows for the quantification of the simulation skill, which is a step in the right direction. I do, however, agree with the second half of the same sentence, i.e., that it is "a convincing demonstration of the effectiveness of [..] integrating [..] models with [..] data".
Citation: https://doi.org/10.5194/egusphere-2024-544-RC1 -
RC2: 'Comment on egusphere-2024-544', Ethan Lee, 10 Apr 2024
The manuscript here presents the modelling of Snežnik, within the Dinaric Mountains in Slovenia, with the aim to present the likely climatic conditions glaciers existed under within the LGM. They use previously studied geomorphological evidence to determine, out of differing climate modification combinations, the 'best-fit model'. This study also takes a look into differing model inputs of climate forcing that can effect the model output.
Overall, the aim of determining the potential climate conditions is, in my opinion, not entirely reconciled in this study. This is due to there being poor constrains (i.e., no palaeorecords) on regional palaeotemperature, or palaeoprecipitation, presented by this study at the time of the LGM glaciation. The use of climatic offsets in the model do however, present a 'envelope' of climate conditions with differing precipitation and temperature offsets that, in combination, could generate ice at the geomorphologically constrained LGM extent.
Some major remarks from my review of the manuscript are:
- As stated above, the determination of the climate conditions cannot rest on singular values of precipitation and temperature offsets. As mentioned in the text, "[Line 445] any increase in precipitation can be countered by a decrease in temperature to keep the conditions for simulated glacier formation about the same". A climatic envelope is more realistic as there is still a large amount of uncertainty in the ice reconstruction, the palaeoclimate, and within the model itself. Further, a climate envelope allows for there to be differing variations in the precipitation and temperatures, that may see palaeorecords fall within the stated range. Following on, a statement on what records could be used, or are needed, to allow a more constrained climate estimate would allow future studies to understand where the gaps are.
- While there has been an attempt to understand how the reconstructed ice is influenced by using differing climatic inputs (temperature forcing using lapse-rate model or WorldClim, precipitation forcing using single value or WorldClim), there is limited expansion on the model sensitivity. Certain model parameters will shift what climatic offsets are needed, and that may change certain numerical outcomes of the study. Within the PDD model for example, Degree Day Factors (DDFs) are likely to cause substantive differences in ice generation, as well as the refreezing factor. Other glaciological physical parameters (enhancement factors etc.), and till parameters (till water content etc.) may also effect the output when using the same temperature and precipitation offsets. I would expect there to have been some consideration of the model sensitivity to certain unconstrained parameters. If this has been done by previous studies within the same region, it needs to be stated within the text. I do understand that this follows off the back of Žebre et al. (2021) and Candas et al. (2020), that have done some sensitivity analysis previously, thus a more explicit indication of sensitivity of the model is needed.
- There is clearly a large amount of uncertainty in the geomorphological evidence, and the geochronology of the region. I did not see any statement on the timing of the LGM, nor how it is known that these are LGM specifically? The studies cited to present the geochronology seemingly look at younger glaciation during the Younger Dryas. Žebre et al. (2019) does state that an age of 18.7 ± 1.0 cal ka BP was found from bone fragments in an outwash, but this is a singular piece evidence for the LGM. This does not definitively constrain the evidence as LGM specifically. The moraines used to constrain the model could be older/younger advances, which needs to have some appreciation in the introduction.
- Similarly to the above comment on the geomorphology used, the model reconstructs ice within the mountain range to the southeast (within the Gorski Kotar Ice Field), where no geomorphology has been presented (at least shown in this study). Žebre et al. (2016) does show (in their Figure 1) that ice was present on this high ground, with a question of ice filling a gap between two ice extensions (or where the ice is reconstructed in the model here). While it is understandable the domain is limited to the Snežnik icefield for computational reasons, a paragraph on the glaciation to the southeast would be beatifical, to present to the wider community where ice is being built, and where future studies should look for glacial evidence. This would also aid in locations which are loosely constrained (dashed lines in this study's Figure 1). Could the model aid in providing a likely area on where to look for evidence to more better constrain the LGM extent?
- Lastly, "[Line 471] This research has successfully established a quantitative framework for the assessment of palaeoglacial simulations that integrates both definitive and provisional geomorphological deduced ice boundaries to improve the accuracy of model results". I agree this is one of the main features of the study to I believe warrants further consideration and development. Are there any major shortfalls that should be noted? How can this be improved upon for future uses?
Below are minor remarks on the manuscript:
- Line 2: State the study location name within the abstract within the location of the Dinaric Mountains.
- Line 26: What is the timing of the LGM for this region?
- Line 36: 'extend' to 'extent'
- Line 65: Citation issue I believe, ARSO needs a date.
- Line 80: Superfluous information. Just state which version of PISM you used, and what it is briefly. Do not need to know what it was run on, unless on a HPC, then in acknowledgments state the HPC and ownership.
- Line 81: The study uses the same parameters are another study with the same author, how far away is the Žebre et al. (2021) study compared to this study?
- Line 83: I am surprised that 'most parameters were .. left at their default values.' A table similar to that sin Candas et al. (2020) would be good, but I do understand there are a lot of tables that could also be combined.
- Line 84: 'e.i.,' to 'i.e.'
- Table 1: You use the -surface_pdd in PISM, but what DDF values were used?
- Table 2: The wind direction (150°) is not inline with what is stated in the 'PISM option'
- Line 89: 'maps' to 'models' as DEM stands for 'digital elevation model'
- Line 137: 'illustration' to 'illustrate'
- Figure 2: Placing the lines of the geomorphology would help the reader understand why certain regions are forbidden and others are not.
- Line 189: I do not understand the sentence ...'precipitation model output is multiplied by a factor to either increase or decrease
the precipitation linearly by several percent to several ten percent'. However later you say on line 357 "...with air temperature spacing by 0.5 °C and precipitation spacing of 10% are shown". Is this a different test or climate offset you are using from that stated before, or the same? Having a varying percentage difference for precipitation makes it confusing to know what percentage change you use for which precipitation offset. - Figure 3: Figure caption require more information. Where is the data from? Elevation of the AWS? What is the period this is for?
- Figure 5: Put what the precipitation model is above the figure boxs, similar to Figure 4.
- Line 244: What interpolation technique was used?
- Line 261: Same as above
- Figure 6: Second line down 'hte' ???
- 4 Results: If there is no discussion section, name this as 'Results and Discussion'.
- Line 357: ...'air temperature spacing by 0.5°C and precipitation spacing of 10%' - The spacing of temperature is correct. But the spacing of the precipitation is not on Figure 8. You say it is a 10% spacing from the 2041 mm baseline. If this was correct, you would be adding 204.1 each time. However, between 2041 and 2320 is 279 (13.7%), while between 2320 (the first column) and 2610 (the second column) is 290 being 14.2% of 2041 or 12.5% of 2320. If the starting point is correct, please correct what you actually used as a percentage spacing for your precipitation offsets.
- Line 358: You say you start the temperature and precipitation offset at -6°C and 2041 mm/a respectively, but do not show it on Figure 8. If you do not show it, state the values as the first model run shown with the lowest numbers used.
- Figure 8: Due to how far the figure is from the initial figure on the colours and their meaning, place them on the figure so we know what they mean. Further, you use °C and K interchangeably, stick with one or the other.
- Line 360: 'An decrease in temperature of 0.5°K coupled with a 10% decrease in precipitation does not significantly alter the extent of ice field.' - I think there is something wrong here, and that you mean, a increase of 0.5°C, coupled with a 10% decrease in precipitation, or vice versa, does not significantly alter the ice field extent.
- Figure 9: While I think I know what is going on, it is not well explained. From the figure, each line colour represents a resolution, is it that they were not resampled past their resolution? So they all start at 400m, and the redline is if it was never resampled, while the green line is when it is resampled to 200 m at 1500 and then not beyond that (i.e., not resampled to 100 m at 2500)? Needs to be better explained.
- Line 413: "The larger ice field located to the southeast is partially responsible, which covers the largest part of forbidden area in the simulation" - Is this forbidden because it is known there was no ice there? Žebre et al. (2019) shows ice was in the Gorski Kotar Ive Field. If it is of a similar elevation and close to the study area here, why would it not have had ice during the LGM?
- Line 422: Rather then the overall 'climate conditions' I believe that it is more of a climatic envelope under which these glaciers can exist under in this region. As there is no direct control over the temperature or precipitation in the region from the palaeorecord, it is impossible to definitively determine the climate conditions needed.
- Figure 14: Addition of a colour ramp for the ice thickness, as elevation for the DEM is the only one shown.
- Line 474: ...'produced an ice field that is consistent with empirical geomorphological reconstructions.' - While the 'optimal' simulation is closest to the geomorphological evidence, I do not think it can be said that it is consistent with the evidence, as there are areas where it is not. Maybe saying that it falls the closest to the geomorphological evidence, and maybe state where there are still regions that cannot be reconciled here?
- Figure 16: Seems there is the southeastern region that does not get entirely covered by the model ever - would be good to see a small sentence or section just recognising it and musing why this may be.
- Lines 487: 'The study reveals that air temperature and precipitation are closely linked when it comes to the size of glaciers.' - This is a well known relationship in models, I do not think it is really the most important result from this study.
- Line 495: Potentially provide how the model shows uncertainty on some of the geomorphological evidence. The model could be used to provide areas that need to be looked at in further detail.
I have attached a PDF that has the locations of these minor remark to aid in editing.
Overall, I believe the work can use some tidying up, with further detail on, 1) certain sources of information that the study is resting itself on (geochronology, geomorphology etc.), 2) the model sensitivity, as this can substantially change the interpretation if different values are used for certain unconstrained parameters, and 3) that some form of track change is needed to present these as an envelope of climate combinations that allow ice to sit at the likely LGM extent here. I look forward to seeing more research from these authors in the future.
Citation: https://doi.org/10.5194/egusphere-2024-544-RC2 - RC3: 'Commented PDF', Ethan Lee, 10 Apr 2024
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
216 | 60 | 13 | 289 | 9 | 5 |
- HTML: 216
- PDF: 60
- XML: 13
- Total: 289
- BibTeX: 9
- EndNote: 5
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1