Icebergs #19
Replies: 7 comments 20 replies
-
HeatLatent heatLatent heat: I suggest deal with it at runtime. A data set could include the latent heat necessary to melt that volume of freshwater, but calculating that is trivial, and I think the ocean model will need to do the heavy lifting to properly incorporate that value. Sensible heatThere are two components to this. Warming ice to the phase transition temperature (PTT), and then warming the freshwater from the PTT (may be up to -2) to the ambient ocean temperature. This may be up to 15 % of the heat budget so I think they should be handled. Ice warming needs to be assumed and parameterized by an ocean model. Warming the meltwater to ambient may be explicit or implicit in how the freshwater forcing is handled depending on the model?
Order of magnitude, from pint import UnitRegistry
u = UnitRegistry()
c_i = 2093 * (u.J / (u.kg * u.K)) # specific heat capacity of ice
L = 334E3 * (u.J / u.kg) # latent heat of fusion for water
k = 4180 * (u.J / (u.kg * u.K)) # specific heat capacity of water
m = 1 * u.kg # sample mass
T_0 = (273.15 - 2) * u.K # phase transition temperature
T_i = (273.15 - 20) * u.K # ice temperature @tyler_2013
T_w = T_0 + (4 * u.K) # ocean temperature @webber_2017
Q_warm_ice = (m * c_i * (T_0 - T_i))
Q_melt_ice = (m * L)
Q_warm_water = (m * k * (T_w - T_0))
Q_tot = Q_warm_ice + Q_warm_water + Q_melt_ice
print(f"Warm ice from {T_i.to(u.degC)} to {T_0.to(u.degC)}: {Q_warm_ice} ({(Q_warm_ice/Q_tot).to('%').round(1)})")
print(f"Melt ice: {Q_melt_ice} ({(Q_melt_ice/Q_tot).to('%').round(1)})")
print(f"Warm water from {T_0.to(u.degC)} to {T_w.to(u.degC)}: {Q_warm_water} ({(Q_warm_water/Q_tot).to('%').round(1)})"): Warm ice from -20.0 degree_Celsius to -2.0 degree_Celsius: 37674.0 joule (9.7 percent)
: Melt ice: 334000.0 joule (86.0 percent)
: Warm water from -2.0 degree_Celsius to 2.0 degree_Celsius: 16720.0 joule (4.3 percent)File formatThe modelers should drive the final format and details (CSV, NetCDF (grid or vector), approximate resolution if gridded, map projection, etc.), assuming there is even consensus between models. Subglacial discharge (surface and subglacial melt) and sub-shelf dischargeFor liquid discharge (sub-shelf melt and surface runoff) I think both Frank and Gavin are suggesting vector format, not gridded. I agree. See comments below on binning by sector/region. IcebergsI think solid discharge should be provide as point-source (per glacier, or perhaps per region) because some model oceans may want to route and melt their own icebergs. We can also provide a spatial map of where icebergs are expected to melt, but I'd like input from others on how to produce this map. Maps of iceberg locations exist (see https://www.scp.byu.edu/data/iceberg/A68tracking.html ) but this is a form of survivorship bias. They melt elsewhere. Elsewhere is not as simple as "icebergs flow north", so icebergs in a given sector may be sourced from other longitudes. I'm not sure what to do about this, and suggest a separate topic to discuss freshwater from icebergs (sources, end-points, melt rate, melt depth). Icebergs (like melting shelf ice, see above) should probably be at -10 or -20 C, unless Peninsula glaciers have warmer temperature profiles. Spatial resolution (of vector data)I suggest the data providers should generate the highest fidelity product they can (assuming higher temporal and spatial resolution does not impose undue effort burden on the work). The benefit of producing the data with higher fidelity is more people will use it outside of this project/community. I have a lot of observationalists and process study people using products that provide glacier- or fjord- scale spatial resolution, but it cost me no effort relative to e.g. 'Zwally sector' which might be all that a GCM wants or needs. The method to bridge between high-fidelity data and model input is a) choosing the right format b) extensive use of METADATA, and c) example code (or 'downstream products'). As one example, my 'Greenland freshwater discharge' product (http://doi.org/10.5194/essd-12-2811-2020) provides daily temporal resolution and stream-scale vector (not gridded) spatial resolution. There are 29430 'stations' (stream outlets). I assume this is worthless as model input. However, for every station I provide the following metadata:
These 29430 stations at daily resolution can then be collapsed to annual runoff per Zwally basin (as exampled by Frank) with just a few lines of code. Data producers can do that as part of this effort, or not, if final use cases are TBD. When someone asked me how make a downstream product, in addition to doing it for them, I document it in the README or on the GitHub discussion page for that product (See README at https://github.com/GEUS-Glaciology-and-Climate/freshwater/ and reduce-by-ROI example at GEUS-Glaciology-and-Climate/freshwater#34) |
Beta Was this translation helpful? Give feedback.
-
|
I like this idea. I'm wondering if we could generalize it further and create related release patterns for longitude bins. In these cases, we could determine the related release patterns for different drainage basins without defining these basins upfront (in detail). This would allow us to make it consistent with basins used in the PICO model (https://dx.doi.org/10.5194/tc-12-1969-2018) to determine basal melting rates in PISM, for instance. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @crodehacke GitHub only allows 1 level of reply-depth so I'm not sure what you refer to when you say "I like this idea" - there are a few ideas discussed above. By 'release pattern' do you mean meltwater release? Or iceberg release? Maybe it doesn't matter. Ideally meltwater, but if all we have is obs "berg position", then work with that. My interpretation of 'release pattern by longitude bin (w/o defining basins upfront)' is to do what we've done above (obs or model), but instead of having 18 (or 10) IMBIE basins, have, say, 360 "basins", one per degree of longitude. I like this idea too. As I wrote above, include metadata for each of these bins (IMBIE basin, Mouginot basin, shelf name, etc.) to make it easy for yet-to-be-determined downstream use-cases to be able to easily work with the data. It would be easy to resample to 18 basins, or 10, or whatever. This is one line of code thanks to xarray, if the metadata is there. What resolution? Could be native model resolution, although a 1 minute ocean model would then generate 21600 maps. Most cells would be 0 and it would compress well... |
Beta Was this translation helpful? Give feedback.
-
|
We discussed the resolution of the maps, and given the necessary smoothing and the imprecision of the modeling, I'm not persuaded that anything better than 1º or maybe 0.5º is really warranted. |
Beta Was this translation helpful? Give feedback.
-
|
I haven't dreamed of archiving such a high resolution of 1° or even finer. I consider several degrees a good choice for reasons listed by @gavinny, for instance. |
Beta Was this translation helpful? Give feedback.
-
|
I've spent a day dividing Antarctic iceberg by 1 ° longitude bins, using the first observed position of icebergs in the BYU DB, and generating 360 maps (1 per initial location). This fails in some locations because Antarctica isn't a circle. The same initial longitude represents different IMBIE basins around the peninsula and near the Western edge of the Ross shelf and Victoria Land. This type of organization is even more complex in Greenland. For now, I think I'll return to dividing by Rignot (IMBIE) basins. FYI, here's 10 random lines from a CSV containing all iceberg observation positions summed by days-in-1°x1°-bin: And an initial NetCDF might look like this: But as I said, I think I'll return to IMBIE regions, so instead of the first dimension being |
Beta Was this translation helpful? Give feedback.
-
|
Yes. This is what I thought we’d agreed to do. However, I’m a little confused as to the number of IMBIE basins we are planning for. There are 17 on the IMBIE webpage, but maybe 10 is better – presumably this just amalgamates some of the smaller ones. We will need to smooth these fields a little for implementation within coarser resolution OGCMs but I don’t think that is problematic. Happy to work with whatever we have available.
Gavin
…--
Gavin A. Schmidt
Director, NASA Goddard Institute for Space Studies
2880 Broadway, New York NY 10025
Cell: 212 749 0006
Email: ***@***.******@***.***>
URL: https://www.giss.nasa.gov/staff/gschmidt
From: Nicolas C. Jourdain ***@***.***>
Date: Tuesday, April 30, 2024 at 10:21 AM
To: NASA-GISS/freshwater-forcing-workshop ***@***.***>
Cc: Schmidt, Gavin A. (GISS-6110) ***@***.***>, Mention ***@***.***>
Subject: [EXTERNAL] [BULK] Re: [NASA-GISS/freshwater-forcing-workshop] File formats (Discussion #19)
CAUTION: This email originated from outside of NASA. Please take care when clicking links or opening attachments. Use the "Report Message" button to report suspicious messages to the NASA SOC.
Hi Thomas,
I agree this is the right approach. Nacho Merino actually coded that in NEMO during his PhD (for 10 sectors as well but not defined based on IMBIE basins), and I used it in my 2019 update of his dataset. Here is the kind of melt pattern that we have (the rectangle shows the origin of icebergs):
Here for the mean melt rate of icebergs from the Eastern Amundsen Sea :
Screenshot.2024-04-30.at.16.07.23.png (view on web)<https://github.com/NASA-GISS/freshwater-forcing-workshop/assets/25052687/976a536d-3d36-49b0-a77a-274266ac7fb4>
Here for the mean melt rate of icebergs from the Larsen & Filchner Ronne area :
Screenshot.2024-04-30.at.16.10.18.png (view on web)<https://github.com/NASA-GISS/freshwater-forcing-workshop/assets/25052687/190200d8-e6e1-417a-a580-3ec11a16bee9>
I think it would be better to redo this with the actual IMBIE basins and with a more realistic coastline (this one is cut at 77°S), but this one is available now if needed.
—
Reply to this email directly, view it on GitHub<#19 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AF2UIXF4GORWHLUYQ6NTIH3Y76SH7AVCNFSM6AAAAABFRWGC66VHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TENZWGEZDG>.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
There are 18 basins. The IMBIE page says
The graphic at http://imbie.org/imbie-3/drainage-basins/ has 18: 10 east, 5 west, 3 peninsula. The data actually contains 19. There is one for "Islands". |
Beta Was this translation helpful? Give feedback.
-
|
In ISMIP6, we decided to combine basins that divide up ice shelves. We ended up with 16 if I recall correctly. It looks likely we would want to combine the Filchner and Ronne basins for purposes of generating icebergs and also the eastern and western Ross. @nicojourdain, what are your thoughts here? |
Beta Was this translation helpful? Give feedback.
-
|
Anna Olivé-Abello, working with Pierre Mathiot and myself at IGE, has actually already estimated the melt patterns from the 18 IMBIE basins. We can probably average over a longer period to have a smoother pattern. This is not perfect as this is a circum-Antarctic domain cut at 55°S, so we are missing ~70Gt/yr of melt in the South Atlantic, but this is probably the best we have in our group without having to wait for new simulations (and we can combine previous patterns to fix the South Atlantic leakage if needed). |
Beta Was this translation helpful? Give feedback.
-
|
Should we be using observations (that provide iceberg density and time, but not melt), or model data? Respond in comments or just vote by clicking on 👀 for Obs, or 👍🏿 for Model. |
Beta Was this translation helpful? Give feedback.
-
|
model for completeness, obs for sanity checking. |
Beta Was this translation helpful? Give feedback.
-
|
@xylar : in ISMIP6, we merged Ross and Filchner-Ronne because we wanted to calculate ice-shelf basal melting in a consistent way under these ice shelves, but for this work, there is probably no need to merge and we can keep 18 basins (I am not sure it will make a big difference anyway). @mankoff : my question about the satellite data you have used is whether they represent only the largest icebergs or all the iceberg population? (I do not have time to go through the papers right now). What proportion of the total iceberg mass is represented. If only the largest icebergs are represented, this could give a slightly different melt pattern even though most of the mass is in the largest ones. |
Beta Was this translation helpful? Give feedback.
-
|
@nicojourdain The BYU data comes from the US National Ice Center (NIC) which states, So... missing all the small bergs. If we have caling (from velocity and shelf front position) and tracked new icebergs (from the BYU obs), the difference are all the small bergs. These could, perhaps, be presumed to melt at the surface and immediately adjacent to the shelf (i.e., at their source). There is another other observational datasets: https://scar.org/library-data/data/iceberg-database |
Beta Was this translation helpful? Give feedback.
-
But we're not (only) interested in iceberg mass. We want iceberg meltwater flux. The graph above (see #19 (reply in thread) on GitHub) displays different properties than melt which probably scales roughly linearly with surface area (not quite due to different depths). In the worst case scenario where surface area is minimized as icebergs break apart and get smaller, an iceberg 10 % of the original size (1 order of magnitude decrease per the columns in the figure) leads to 22 % of the surface area. This means 22 % of the original melt even though the new iceberg is 10 % of the original size. It's possible to have iceberg B with an order of magnitude less mass than iceberg A, but still have surface area of B > 0.22 * surface area of A, meaning higher melt rates. Worst case scenario 2 order of magnitude mass decrease = 5 % of original melt. It could be much larger if surface area is larger. The graph above also doesn't provide # of icebergs, just %, so I'm still not sure how to map the above numbers to that graphic, or other data from Tournadre (2016) http://doi.org/10.1002/2015jc011178 |
Beta Was this translation helpful? Give feedback.
-
|
Following up... I think that a 4 order of magnitude reduction in mass leads to (worst case) 0.2 % of the surface area (melt). From the above figure, looking at red (# of bergs), left column is ~100 %. Right column is ~1 %. That means melt from the left column (smallest) is approximately equal to melt from right column (largest). Am I doing something wrong here? Happy to share my work (code). |
Beta Was this translation helpful? Give feedback.
-
|
My definition of 'worst case' is incorrect. I assumed an initial square berg (minimum surface area to volume ratio for a six-sided surface (e.g., not a sphere)), and then 'worst case' meant the smaller berg was also square. In reality, the initial berg is not likely to be square -> increased melt per unit volume. Smaller bergs could therefore have a lower area : volume ratio (more square) which would decrease their (relative) contribution, or a higher area : volume ratio which would increase their relative contribution. But keeping things square (or spherical), it seems that all the small bergs may contribute roughly equal meltwater to the large bergs. I think... I think this means we need to handle this - not ignore it. How it is handled is a function of how the iceberg product is derived. For example,
|
Beta Was this translation helpful? Give feedback.
-
|
Ah. Sorry for working this out in public, but I realized another limitation of that figure. It doesn't capture residence time (which I was getting at above with mention of melt rate). The 1 % (by number, 70 % by area) big iceberg(s) may last a decade. The 75 % (by number, 1 % by area) small icebergs may last < 1 year, so > 10x as many per decade, in which case, again, the small ones could matter. Anyway, I think more work is needed here, or someone else can offer an interpretation of this figure. |
Beta Was this translation helpful? Give feedback.
-
|
sorry my label was wrong, black was iceberg area, not mass, I've corrected the label in my previous post. You are right, melt does not scale with the iceberg volume, it partly scales with the iceberg perimeter times thickness (lateral melt due to convective plumes and waves) and partly with the iceberg area (basal melt). The total iceberg area mostly consists of large icebergs. In terms of perimeter, an equal length would be represented by each bin of the histogram. This can be obtained by integrating area or perimeter weighted by Tournadre's power low. I view these distributions as a snapshot of the iceberg population, I don't think it tells anything about the fate of individual icebergs. |
Beta Was this translation helpful? Give feedback.
-
|
Yes – I think we should just use this, but do we need to think about the conversion to total melt so that we can derive a weighting function? I don’t think that we need to worry too much about the small missing flux compared to the total.
Gavin
…--
Gavin A. Schmidt
Director, NASA Goddard Institute for Space Studies
2880 Broadway, New York NY 10025
Cell: 212 749 0006
Email: ***@***.******@***.***>
URL: https://www.giss.nasa.gov/staff/gschmidt
From: Nicolas C. Jourdain ***@***.***>
Date: Tuesday, April 30, 2024 at 12:22 PM
To: NASA-GISS/freshwater-forcing-workshop ***@***.***>
Cc: Schmidt, Gavin A. (GISS-6110) ***@***.***>, Mention ***@***.***>
Subject: [EXTERNAL] [BULK] Re: [NASA-GISS/freshwater-forcing-workshop] Icebergs (Discussion #19)
CAUTION: This email originated from outside of NASA. Please take care when clicking links or opening attachments. Use the "Report Message" button to report suspicious messages to the NASA SOC.
Anna Olivé-Abello, working with Pierre Mathiot and myself at IGE, has actually already estimated the melt patterns from the 18 IMBIE basins. We can probably average over a longer period to have a smoother pattern.
MAP_NEW1v1_BedMachine_meltingbasin_lr.png (view on web)<https://github.com/NASA-GISS/freshwater-forcing-workshop/assets/25052687/810833f0-0a03-4844-8895-d25cd05bffbd>
This is not perfect as this is a circum-Antarctic domain cut at 55°S, so we are missing ~70Gt/yr of melt in the South Atlantic, but this is probably the best we have in our group without having to wait for new simulations (and we can combine previous patterns to fix the South Atlantic leakage if needed).
—
Reply to this email directly, view it on GitHub<#19 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AF2UIXFV7ZXYNRVJCYBWYIDY77AKPAVCNFSM6AAAAABFRWGC66VHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TENZXGU4TO>.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
Noting this new publication for group awareness: https://gmd.copernicus.org/articles/17/3279/2024/ |
Beta Was this translation helpful? Give feedback.
-
|
Adding @ackerlar here in case he has further input |
Beta Was this translation helpful? Give feedback.
-
|
Thanks to introductions from the SOFIA group I've found @julianamarson, their paper 2024 http://doi.org/10.1029/2023jc020697 and the associated data at https://canwin-datahub.ad.umanitoba.ca/data/dataset/nemo-anha4-seaice-locking-icebergs/resource/8aa9c193-214e-4152-9abe-037010bf1999 This is still a work in progress but I think we now have a Greenland product complementing the graphic from @nicojourdain above. That is, meltwater from icebergs distributed around Greenland and tied back to a source location. Work is ongoing at the |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
From Ben Davison:
From @gavinny
From Frank Pattyn:
Beta Was this translation helpful? Give feedback.
All reactions