Of sea-ice models and phytoplankton.

00:00:02: A hoi and welcome to the Gen Z podcast, a podcast by early career marine scientists.

00:00:10: Hi I'm

00:00:11: Leonard And i am currently writing my master thesis To finish up my Master in Environmental Modelling at The University of Oldenburg In Germany!

00:00:20: And hi ,I Am Shalotte !

00:00:22: I have recently graduated from the Masters in Marine & Coastal Systems At The University Of Algarve in Portugal And together we are going to record some episodes for this podcast over the next few weeks.

00:00:34: So, Leonard let's dive right in!

00:00:43: And

00:00:45: my main research question was to find out whether different sea ice models change how a biogeochemical model simulates primary production in the Arctic Ocean.

00:01:09: There were two model simulations available at the Nansen Center.

00:01:15: So, the data was already there and I didn't really have to run the numerical models.

00:01:21: But what we mentioned before it is one of these classic cases where a lot of data are available but nobody looks into them.

00:01:31: in this case that's me!

00:01:38: And I had a very cool supervisor team of three amazing female researchers.

00:01:46: So my main supervisor was Dr.

00:01:48: Anette Samuelsen, and my second supervisor was Doctor Heather Reagan.

00:01:54: And i also hide a lot of support from another colleague, Dr.

00:01:58: Shuangau during the process of doing my research.

00:02:14: Sounds Very Good!

00:02:15: Could you maybe shortly explain what primary production is?

00:02:20: Why is it important?

00:02:21: why do you want to look into it.

00:02:24: Primary production, is a process that all green plants on this planet perform.

00:02:30: It's when plants use the energy from sunlight To produce organic carbon so sugars From carbon dioxide.

00:02:41: and This process happening in forests because every tree performs this process which Is called photosynthesis.

00:02:48: But it also happens in the ocean because phytoplankton, which are micro algae.

00:02:55: So also green plants they do the same.

00:02:58: and It is important Because this microalgae the phytoplancton Is first of all The basis Of the marine ecosystem?

00:03:08: Because it is the foundation of the food web.

00:03:12: so We have the microscopic phytoplankton, and then this is eaten by so-called zooplanktons which are also small organisms.

00:03:22: And these are again eaten by larger organisms such as fish... ...and then fish of course are eaten by orcas which are apex predators.

00:03:31: So yeah it's a foundation for food web in the oceans.

00:03:35: But then also phytoplankton and the primary production they perform is very important because it helps us to mitigate climate change.

00:03:44: Because, They use carbon dioxide to produce sugars.

00:03:49: so The CO-II we have in the atmosphere parts of it are also entering the ocean And there consumed by microalgae And once this organic carbon is stored away in deeper water layers or even in the sediments of ocean seafloor, then CO₂ is stored for a long time and that's what helps us to take it out from the atmosphere.

00:04:18: Okay, so I see that the ocean and

00:04:29: its algae community is quite important for environmental considerations in a topic of climate change because their carbon starts.

00:04:41: But what's actually the link to the CIS models?

00:04:44: And which context is CIS important here?

00:04:49: That's very good question!

00:04:51: It applies not only to arctic oceans but also through the Antarctic Ocean The sea ice always moderates the exchange processes between the atmosphere and ocean.

00:05:05: And we know that the Arctic Ocean is warming much faster than the rest of our globe, this process called arctic amplification... ...and most drastic estimates actually say that the arctic warms four times faster then the world.

00:05:27: And this also leads to an observed change in how the sea ice behaves.

00:05:33: So, the sea-ice has become and is still becoming much thinner and younger... ...and it's more seasonal.

00:05:41: so In two thousand twenty four for example that was a year where satellite records recorded the sixth lowest maximum sea ice extent.

00:05:53: So, the area that is covered by seas is constantly shrinking and this not only leads to problems for polar bears who don't have any areas of hunting or living anymore but also changes how the algae community in the ocean behaves or grows.

00:06:19: And one of the most straightforward factors here is, of course, light.

00:06:25: Because we've talked about it before that microalgae and green plants in general need light as source energy for producing sugars... ...and if there's very thick sea ice nearly no light will reach the ocean so algae cannot grow in these areas.

00:06:46: However, if the sea ice becomes thinner or even starts to break up then light can reach the sea surface and this is also where the phytoplankton will start growing.

00:06:59: Depending on how much light is available it'll grow more.

00:07:03: depending when the light becomes available it's starting to grow earlier for example.

00:07:10: From the observations, we can see that change in sea ice has already reached to an increase on primary production.

00:07:19: We have less sea ice and therefore more algal growth.

00:07:25: It is also estimated this trend will increase over the next decades with climate change.

00:07:37: I'm not saying it's a good thing of course But it is something to consider in climate models, for example.

00:07:48: Because we know that from this increased primary production the Arctic Ocean... ...the ocean also likely to sequester and store more of the carbon And hopefully will have positive feedback on mitigating climate change.

00:08:02: Therefore its important to consider this when forecasting future.

00:08:07: I see that sounds reasonable.

00:08:15: And one of the big important factors in this changing CIS regime that I mentioned before, so it's becoming thinner and younger is that the CIS will start to break up much more.

00:08:32: So its likely to form openings... ...and these can be smaller openings called leads.

00:08:40: They're at a scale of ten to hundreds of meters.

00:08:44: And then we have larger openings, which are called polinias and they can be tens-to-tens of thousands of square kilometers... ...and often occur every year in the same region.

00:08:57: So these two main types of openings that I will also likely be referring more often when talking about my master thesis

00:09:11: in several... Yes,

00:09:17: absolutely.

00:09:19: But coming back to the eyes that's probably where the difference is in the CIS models comes into play no?

00:09:25: Yes exactly.

00:09:26: and there was a question because we have one CIS model That it better modeling these smaller fractures than openings then other ones.

00:09:38: And this why wanted figure out if This difference in sea ice behavior and the sea ice models has this effect that we are expecting.

00:09:49: That it actually changes light availability and does a primary production, but also to mention here Light is not the only pathway how the Sea Ice will affect primary production.

00:10:03: It can also affect primary Production by having an influence on how water column is mixed.

00:10:11: Basically when you have no sea ice, then

00:10:15: the wind

00:10:15: can reach the sea surface and therefore amplify the mixing of water column.

00:10:19: Whereas if you have sea ice than the

00:10:22: upper

00:10:22: ocean is sheltered from the influence of wind.

00:10:25: so in summer or areas of these fractional features of openings there's

00:10:32: a potential impact

00:10:33: on the wind on the water-column mixing And also when sea ice melts It introduces freshwater

00:10:41: at

00:10:42: the sea surface or into the upper ocean and that leads to something we call shoaling, so a decrease in mixed layer depth.

00:10:53: In the water column... And the opposite happens when the sea ice forms.

00:10:58: So it's freezing because during this process very salty liquid which is called brine is ejected and the salty liquid is much heavier.

00:11:11: It has a higher density, therefore it enhances the watercolumn mixing.

00:11:16: And through these two processes The sea ice influences where in the water column you can find the phytoplankton.

00:11:23: So if the mixed layer is very deep then the phyto-plankton will be in very deep layers as well maybe Where the sun doesn't reach whereas If the mix layer is shouling brings the phyroplanktons to top

00:11:38: to the top where there

00:11:38: is light so it can start growing and producing.

00:11:42: It also influences of course, the distribution of the zooplankton in the water column.

00:11:46: So...it has an effect on balance between biomass producers and biomass consumers And therefore offcourse..on whole amount available biomass in a water.

00:12:00: Then another very important part or aspect that Phytoplankton not only needs light and carbon dioxide, it also needs additional nutrients to perform photosynthesis.

00:12:14: And

00:12:15: during the water column mixing, the nutrients from deeper water layers are brought up into the upper ocean where there is light.

00:12:23: therefore primary production happening

00:12:26: So through this

00:12:28: replenishment of nutrients in the water column, mixing of the water columns and mixed layer depth is also very important for primary production.

00:12:48: Okay so I see that existence of those breaks on the sea ice.

00:12:55: it's quite important for phytoplankton and the primary reduction.

00:12:59: And now you said one of these sea ice models behaves differently.

00:13:05: Can you explain a bit?

00:13:07: Yes, of course.

00:13:08: So the one CIS model I was looking at is called Nexim and it's actually an in-house CIS from the Nansen Center.

00:13:18: And this one is relatively good in modeling these cracks or openings because it models the CIS as a brittle material that breaks much easier under mechanical stress.

00:13:34: And the second simulation that was available, it ran with the Los Alamos sea ice model size.

00:13:44: This one simulates more like a very viscous elastic layer so something like the maple syrup you would put on your breakfast pancakes and this is not good.

00:13:57: in modeling these smaller cracks or openings Yes, and by comparing these two simulations with these two CIS models we were hoping to figure out how this affects the primary production in a biogeochemical model.

00:14:14: I like the comparison of maple syrup or honey.

00:14:19: yes for sure!

00:14:22: I guess that having additional value modeling those cracks there must be some kind of disadvantage one next time.

00:14:31: can you say more about differences?

00:14:35: Yes, I wouldn't call it disadvantages from the NEXEM model.

00:14:44: It relates to what we were talking about in last week's episode on how numerical modelling is very complex and depending upon your purpose of your model you will develop a different model.

00:14:59: so size, the syrup sea ice model.

00:15:02: It just has been around much longer or it has been developed much longer than Nexim, so for example the representation of thermodynamic processes in this sea ice such as melting and ice formation.

00:15:18: they are simulated a bit more complex way which probably also represents reality better.

00:15:31: Okay, so I see the mechanical behavior is what you are really looking into because the thermodynamics Are pretty important for CIS but not in this case For the primary production or not.

00:15:42: What?

00:15:42: You're looking at.

00:15:43: So can you tell me how?

00:15:46: next?

00:15:47: some it's better and this way of doing does a mechanical behavior.

00:15:52: The challenge usually with representing these fractal features is that they are temporally and spatially very variable.

00:16:01: so what I said before leads can be only meters, have to be like five kilometers or even finer.

00:16:24: Otherwise the CIS models will not be able to model these features, Like normal CIS Models on that viscous CIS Model such as size

00:16:34: which is quite a lot.

00:16:37: I mean if you look at for example they the earth system models i think They run of one degree Which Is like A hundred Kilometers.

00:16:45: Exactly.

00:16:47: Yes and this is The exact problem Even If You want To only in quotation marks model the whole Arctic Ocean.

00:16:54: That's also huge, so you cannot model that whole Arctic ocean on a model grid of five kilometers.

00:17:00: it is not possible for computational costs reasons and this was also starting point to researchers at the Nansen Center looking into how to represent mechanical behavior a case where the model can be run on realistic model resolution, which is computationally doable.

00:17:37: So far we had an nice introduction to your topic and what you wanted to ask in your thesis but now could tell me more about how did actually do research?

00:17:52: What were your methods to proceed?

00:17:58: so yes as mentioned before The two model simulations I was working with, they were already available at the Nansen Center.

00:18:08: And it was one simulation of a high-comic Cosmo biogeochemical ocean model that was coupled to the CS Model Nexum and second simulation was High ComiCosmo couple to size.

00:18:22: so... This biogeochemical ocean model setup was modeling the Arctic Ocean physics, biological chemical and geological processes.

00:18:33: And then we had addition of the CIS models there.

00:18:36: Okay so could you maybe give a bit or an example between the physical and biogeochemical process just to be able to distinguish them better?

00:18:46: Of course!

00:18:47: So the ocean physics would be for example the mixed layer depth So how deep the mixing goes in the water column, which also means how thick is the water layer where we have these rather uniform homogenous characteristics.

00:19:08: That would be covered by the Arctic Ocean Physics model and the biogeochemical processes, they are really ones that related to primary production.

00:19:21: That would be for example a simulation of nutrient concentrations in water which is affected by inflow from rivers but also resuspension So how the nutrients are actually moving in the water column, but also like they're moving from the sediments back into their other way around.

00:19:44: And that is all handled by the biogeochemical component of the model setup.

00:19:50: But there's also models as said before The concentrations of phytoplankton and also the concentration of zooplanktons which is consumers of the phytoplancton.

00:20:01: I hope this clarifies it a little bit.

00:20:11: Models they're usually forced and their nested, and there relaxed to certain data.

00:20:18: So basically that describes the external conditions that the model draws from?

00:20:25: And I just want to mention here may be Both simulations were set up in a way that they are as comparable possible and really the only difference between those two simulations was coming from the sea ice models.

00:20:43: One thing actually made a difference also then looking at results is that size sea ice model, so the viscous one did allow light to go through thin eyes whereas the Nexem sea ice model does not allow light to go through ice at all.

00:21:02: So if you have an area on your grid that is ice-covered, there's no going to be a light in the ocean and nexem where as it has an areas of ice covered in size If the ice is thin enough Light will actually reach the ocean.

00:21:16: And maybe just like with this little spoiler here we did see of course that this affected the primary production In comparison between these two models?

00:21:26: Yeah, and both simulations were available for a ten-year time period.

00:21:33: So the simulation period was two thousand seven to two thousand sixteen And we had daily data.

00:21:39: so that was a lot of data For ten years.

00:21:42: if you have like three hundred sixty data pieces per year

00:21:48: See just out of interest how long does it take to run those models for this period of ten years?

00:21:55: just to see how much computational effort is actually behind this simulation.

00:22:00: Do you know that?

00:22:05: It depends a little bit on the model setup, and I can't tell you exactly how long it took to run these simulations because they already had happened once i joined Nansen Center.

00:22:16: However, and then it also depends for example on which super computer you run.

00:22:20: It's like how fast this one is.

00:22:22: And Yeah some circums like y'know?

00:22:26: I'm not so how many other colleagues are running things under supercomputer at this point.

00:22:31: yeah but i think For a ten year model simulation run Like This.

00:22:38: I Think it took three to four weeks To Run it the computational efforts and how much time you need to also process data.

00:22:55: At some point, for like more detailed analysis we decided just choose two target years.

00:23:02: And we chose, in the year of three, to look into more detail because it was a recorded minimum sea ice year and I was looking at three different regions of the Arctic.

00:23:16: so i'm looking for the Barents Sea which is an area where you have an Ice Edge.

00:23:21: then i looked at the Arctic Atlantic thick and very compact sea ice.

00:23:33: And then I looked at Chakchi Sea, which is on the Pacific side of the Arctic... This was interesting because it was right at the Bering Strait or where the Pacific water flows into the Arctic

00:23:45: Ocean.".

00:23:58: What did you find in those different characterises?

00:24:05: Well, let me preface this here.

00:24:08: I did find a lot of small and sometimes very funny differences between the two simulations And...I also have to say that in the end i was not able to explain all of the stuff I saw in data.

00:24:29: In some cases, i was not able to trace back what model did and how it reached the data output that I saw!

00:24:38: And also... Of course I will not present all findings here now but I'll try focus on more major findings.

00:24:51: So first of all the model setups, both were really good in simulating the difference between the low CIC or year two thousand twelve compared to two thousand and thirteen.

00:25:05: We did see that decrease in CI's thickness also this decrease in CS concentrations moving from two thousand twelfth to twenty thirteen.

00:25:15: then Overall, it was interesting to see that the CIS concentration in nexum so the brittle CIS model Was lower than two out of three regions.

00:25:28: That I analyzed when comparing into size.

00:25:31: We also saw that the decline and CIS Concentration in spring So once the warming starts It's also stronger in Nexum compared to size In all Three Regions.

00:25:43: And that was something I don't want to say we expect it, but uh...we could explain relatively easily because this brittle behavior of the sea ice and breaking up more openings in these seas as mentioned before.

00:25:58: These openings allow for energy to enter the ocean so that the ocean starts to be warmer in nexum, and therefore it enhances the melt of surrounding sea ice.

00:26:09: And this is likely one reason why the sea ice concentration decline was more rapid at nexium compared to size.

00:26:24: So we expected that light would be the main determinant for spring bloom when primary production starts.

00:26:32: This was the case with all regions.

00:26:34: in both simulations.

00:26:36: We really saw that once light started to become available, the biomass production just increased.

00:26:42: What was interesting to see is that CIS concentration in some cases was a most important predictor of light availability even though in size.

00:26:55: remember the light could go through thin ice.

00:27:00: So you would expect CIS thickness to be more important?

00:27:04: In the case of size am I getting it right?

00:27:06: ?

00:27:08: Not necessarily that CIS thickness would be more important as a predictor for biomass production, but even though there was light going through ice in size the primary production still larger in nexum.

00:27:27: The importance of the lights going through thinner ice and size appeared to more relevant in areas of overall lower sea ice concentrations, and the light through ice was less relevant to very high sea-ice concentration than very high Sea Ice thicknesses.

00:27:51: So much for the light availability as a main factor for primary production.

00:27:55: but I mentioned before that we also have the sea ice effects on mixing with watercolumn And that was something we found, that NEXEM always had a deeper winter mixing in all of the regions.

00:28:14: Supposedly this can be related to ice not being such good shield from wind and atmosphere.

00:28:23: so there is actually energy input through cracks leads into the ocean which enhances the mixing In one region, I actually also found that the difference in mixing could be related to the sea ice thickness and ice formation processes.

00:28:44: So I saw it at the Arctic Atlantic Region where I was analyzing that the sea-ice thickness increased much more in nexum compared to size And therefore... The assumption is that a lot of this very salty water from the brine was injected into the upper ocean, which enhanced mixing in water columns.

00:29:08: So that's also interesting to see and maybe but this is something a little bit difficult to tell there could have been some influence of meltwater in some regions.

00:29:29: Sholing of the mixed layer was associated with the meltwater input.

00:29:40: Very interesting results so far already, but before you spoke about like funny differences we found maybe can say a bit more about this and why they were funny or surprising to you?

00:29:52: Yes for sure I'd like to talk about because i really think it's funny!

00:29:58: So in the ChuckGC that I was analyzing as region huge peak of primary production.

00:30:07: You will see some when around March, usually April was just not showing up so we had the expected peak production in two thousand and twelve in both simulations but then in twenty thirteen only size showed a peak production And Nexim Just didn't show that curve like it showed A very flat curve for Primary Production could be related to like different factors.

00:30:37: For example, maybe in this simulation there was extraordinary growth of zooplankton that consumed a lot of phytoplanktons and therefore the peak doesn't show.

00:30:49: or maybe the composition of the phytoplankton community changed so that available nutrients were not suitable anymore?

00:30:59: Yeah I guess it's definitely the opposite of what you expected from this simulation?

00:31:06: Absolutely.

00:31:09: It's very unexpected and I really cannot trace back or understand why the model did it this way, yes but its something that i'm still working on.

00:31:17: so stay tuned everyone.

00:31:19: maybe we'll find out where the peak is missing.

00:31:36: Definitely an intriguing question!

00:31:39: So...I hope that you will find your pick.

00:31:42: But to finish up Maybe they have some conclusions that you would tell us in the end.

00:31:53: Yes, of course!

00:31:54: The main conclusion is and this was nice to find... ...that the difference in CS models did actually show effects on... the environmental conditions that were then showing differences in timing and magnitude of biomass production.

00:32:13: So, there was one great case where we started out with an assumption for our research question... ...and also say that assumptions are at least broadly speaking.

00:32:28: correct and that the simulations, the models fulfilled our expectations.

00:32:40: So looking at your results on their conclusion in the end which model would you suggest using for future research regarding primary production?

00:32:49: And maybe also prediction of

00:32:51: primary production?

00:32:52: or

00:32:53: yeah may be just elites who knows?

00:32:59: That's not easy to answer because I think overall both of the sea ice models in the couple systems.

00:33:08: they were very good in representing the real world.

00:33:13: And I also compared, in the end... ...the primary production rates that both of these simulations showed with observations that were available from some of the regions that I analyzed and there was differences between them as well a little bit at the onset time.

00:33:35: But overall, both simulations represented the real world and were fit for the purpose of modeling primary production in the Arctic Ocean.

00:33:47: Also both of these models are constantly under improvement and development.

00:33:52: It would, of course make sense also to improve comparability between the simulations if the NEXEM simulation will be run again with allowing light through the eyes as it already does in the size simulation because this factor especially seemed relevant for on-site of the spring bloom.

00:34:13: so that's interesting.

00:34:14: what happens at NEXem if the conditions are the same as in size.

00:34:21: But yeah, maybe I can try to answer your question a little bit more general way by saying that primary production in the Arctic Ocean and that they are definitely relevant for predictions of climate system or in a climate system.

00:34:50: And yeah, because it just ensures that the sea ice simulation is closer to real-world too.

00:34:59: therefore makes sense to consider them.

00:35:04: but yes both models fit.

00:35:08: in my opinion or from how I understand it, both of these models are suitable to model the primary production and arctic.

00:35:29: Thank you very much for this nice summary on your topic so far!

00:35:34: I definitely learned quite a bit because...I haven't worked with primary production especially not in the Arctic but Maybe just as like a bit of roundup, maybe I could ask you what challenges did you encounter in your research?

00:35:53: And was it a bit like the obstacles.

00:35:57: Well for me one of the big challenges i encountered and we talked about this before was the runtime of the simulations.

00:36:06: once...I found myself on the situation that one of those simulations had to be run again and then also the data processing times, because I did not have experience with this before.

00:36:19: And... ...I was surprised to see how long it takes just to process your data that you can actually start to analyze it!

00:36:29: So It Was Also a Lesson in Personal Time Management & Planning In This Case Yes..and Then Also One of the, yeah.

00:36:41: I mean i think this applies to like writing your master thesis in general but uh...I thought it was especially challenging because..yeah But I just didn't have any idea how long would take.

00:36:52: you know or that it would take so long Yeah and sometimes also wasn't really good in doing something in parallel because when processing my data, there was always like some small bugs.

00:37:08: And so you couldn't just have it run on the background and then read some papers for your literature review Because I was looking at my screen hoping that they would not be a bug That stopped the data processing.

00:37:28: But yeah, I mean...I think you encountered similar problems.

00:37:34: A couple of times i would say for sure!

00:37:36: I

00:37:38: also had this definitely and still have it.

00:37:40: so..i think that's a general problem

00:37:44: there.

00:37:50: Then maybe just to conclude everything?

00:37:53: I'd like to ask if you have any recommendations from people who are in maybe a similar stage their life at the moment.

00:38:02: What's writing the most of these?

00:38:04: Do you have any tips that he would like to

00:38:06: share?".

00:38:08: Yes, I do.

00:38:10: Yeah for sure!

00:38:13: Something i found out in the end... That writing down your results and your discussion really early on is very helpful.

00:38:23: I started way too late writing down my thoughts.

00:38:25: I mean, it was always just compiling the graphs of my data analysis and some thoughts in PowerPoint presentations to present them to my supervisors and discuss this with them.

00:38:37: but its not same as when you write down continuous thoughts on results or ideas that we have about.

00:38:47: So I would really recommend to everyone, start early doing this.

00:38:52: Also have your supervisors proofread these chapters or paragraphs earlier on because it will also help them to support you in the analysis and then discussion And It'll also help you too Have a more relaxed submission of your thesis If your supervisors don't have to prove read everything last week before submission deadline.

00:39:16: So that's one big lesson I learned.

00:39:25: And it also applies to your literature review.

00:39:29: I don't think for me, It was necessary To do all my literature review before doing My data analysis and would not have worked Because of the longer Data processing times For example that we just talked about.

00:39:40: But you know...it should be something That You'd rather in first half.

00:39:44: Just understand Your data better as well And helps with interpretation.

00:39:53: Also what i found very helpful get a better understanding of the data and to synthesize it much better was too do make a poster.

00:40:03: And we did this for joining the ICMARE conference, um...and that was very good step from me starting to synthesise my findings.

00:40:12: results I can recommend as well.

00:40:17: Then honestly Just be brave to leave gaps in what you write down and try to figure out because, You will not able explain everything that is in your data.

00:40:30: It's also not the expectation for a master thesis And it isn't the expectation of research in general.

00:40:36: There are always going to open questions That something... ...you really need remind yourself constantly.

00:40:46: Yes, and then I think also important is To not hesitate to ask for help And that applies too general knowledge and background.

00:40:58: But it also applies a very practical aspect such as Python scripts.

00:41:03: So you for example Leonard.

00:41:04: You shared some of your python scripts with me during this process of writing my master thesis?

00:41:10: That was very helpful.

00:41:11: so thank you again for this year.

00:41:15: Oh It's a pleasure

00:41:19: But I think it's also, i had the impression that its like a relatively common thing to do this in my researchers.

00:41:50: Lennart Montag and Schalotte Walter.

00:41:53: Music Florentin Seifert.

00:41:55: Production Graphic Design Annabel von Jakowski.

00:41:59: Management Eva Rolfa.