NCAR Climate and Global Dynamics Laboratory Paleoclimate Advances Webinar Series, 8 Jul 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 4th PAWS Webinar

Description

The fourth webinar from the Paleoclimate Advances Webinar Series (PAWS) which took place on July 8th 2022.

Lily Hahn discussed "Mechanisms of polar-amplified warming" and Eli Tziperman discussed "Cloud feedbacks that may help explain observational constraints on continental Eocene climate"

For more information and to signup for the PAWS Google Group visit:
https://www.cesm.ucar.edu/events/webinars/paws/

A

Okay, now we have recorded okay.

A

Production slide.

B

Can you all say that.

B

Okay, great, let's get started hello. Everyone welcome to the paleoclimate advances webinar series. Before introducing today's speakers, I would like to briefly remind people our code of conduct and the format of the webinar. By joining this webinar, you agree to follow our code of conduct during the webinar. Please consider new ideas, encourage innovation, offer constructive feedback, acknowledge teamwork, show appreciation and share the air. The webinar has two talks. Each talk has 20 minutes with five additional minutes for q and a in the end. We will have about 10 minutes for general discussion.

B

I will give a two minute reminder for each speaker just to keep us on track. Our co-host, professor bronwyn konecki today, will help us moderate the q. A if you have questions for the speakers please either type them in the chat box or use the right hand function. As always, we welcome speaker nomination. You can fill out the nomination form on our webpage or contact anyone on the organizing community.

B

It is my great pleasure to introduce today's speakers with lily khan and the professor elizabeth lily han is a phd candidate at the university of washington, working with professors, dave batisti and the kyle armor.

B

She has done important work summarizing and investigating the mechanisms of polar amplification, including its atmospheric and seasonal asymmetries, she's an nsf graduate student fellow and an arcs scholar, dr eli zipperman, is a professor of oceanography and applied physics at harvard university.

B

He has done important and influential work on the equivalent climates and many other fundamental questions in climate dynamics. Professor zipperman has recently published a book entitled global warming, science, a quantitative introduction to climate change and its consequences. Please check it out if you are interested without further ado. Let's start the webinar with lily han. I will stop screen sharing.

B

Really, please go ahead.

C

Answer, can you see my screen? Okay,.

B

C

Okay, let's hide these well thanks for that introduction, and thanks so much for the opportunity to speak um just a disclaimer that this work has focused mainly on the recent past and future projections for polar climate change. But I'm excited to think more about pilot climate perspectives and hope.

C

There's some interesting overlap so I'll be talking about contributions to polar amplified, warming and its seasonality, um and I'm a phd student at the university of washington with kyle armour and david batisti, and also worked with cece bits, aaron donahoe, ian eisenman and mark zelinka on these projects.

C

So looking first at historical and near surface temperature trends, we can see in these observationally based data sets this pattern of having greater warming in the arctic than in the tropics or in the antarctic, which has had some historical cooling in these datasets um and if we define a polar amplification as the near surface, warming north of 60, divided by the global, mean warming. um This gives a factor of 3.5 in the arctic, for polar amplified, warming and in the antarctic, there's been slower.

C

Warming compared to the the global mean um giving a smaller factor there. So how does this look in the latest generation of climate models here? I'm showing um the ensemble mean of sum of six models, looking at historical, near surface temperature trends, and we see similar patterns of having stronger warming in the arctic compared to the rest of the planet.

C

um Although uh the sensible mean overestimates the warming at all other latitudes in the arctic, compared to the observations, and so as a result, there's uh underestimated um factor polar amplification and this symmetrics ensemble mean compared to the observations um for the arctic and it actually overestimates um antarctic warming compared to the observations but generally getting this. This observed pattern of having um arctic amplified warming and weaker warming in the tropics and antarctic.

C

So how does this compare to intermodal spread and internal variability here, I'm showing in the yellow shading the range of intermodal spread versus map six and then in the dashed orange lines, a measure of internal variability. Looking at um this esm2 large ensemble, two standard deviations plus or minus compared to the zoom of six mean, and so um we can see that these observed warming trends in black largely fall within this range of intermodal spread and internal variability.

C

For most latitudes, except for the um southern hemisphere, mid-latitudes, and so generally, the models are capturing. These patterns of arctic amplified warming compared to weaker warming in the tropics and antarctic.

C

If we look at the seasonal and vertical structure of warming for observations, or this observationally based re-analysis data set compared to sum of six in the arctic, um we can see that the models also generally capture um these patterns of having winter amplified, warming um and also surface amplified. Warming um with weaker warming aloft and in summer there's some interesting differences between the models and observations.

C

So the models are missing. This april peak in warming, that's found in observations, although the um the observations do fall within the intermodal spread and internal variability for a single model for april, but I think that's sort of an open question of why the models underestimate this april warming in the arctic near the surface and the models also have overestimated.

C

Mid-Tropospheric arctic warming compared to observations for this latest generation, and this could be related to having too warm of temperatures in the mid-latitudes, which would lead to um too much heat transport from the mid-latitudes to the um upper atmosphere in the arctic and uh yeah. That could that could support this overestimated um temperatures aloft in the arctic.

C

Another possibility is that there could be too much atmospheric short wave absorption by water vapor in the models in the arctic, um but I think this is also an open question of what controls the vertical structure of warming in the arctic um and has been less studied than the the near surface warming in the arctic.

C

um So, despite some differences, the cmip6 models are generally agreeing with observations um on these patterns of cool amplification and they show these two key asymmetries of having this difference between the poles with greater warming in the arctic than in the antarctic, and also the seasonalized symmetry where the arctic near surface is warming most in winter and least in summer time.

C

So here I'll focus on what drives these patterns of polar amplification and its asymmetries, and I think these asymmetries can be used as a window into the underlying processes supporting arctic amplification and polar warming in general and I'll talk about a couple of papers and first look at what drives polar amplification and this hemispheric asymmetry.

C

So to do that I'll switch to some more idealized experiments with abrupt co2 quadrupling and I'll show averages centered on year, 100 after co2 quadrupling compared to pre-industrial control. Experiments um here we're looking at near surface warming in these experiments um versus latitude, and we can see in symbolic five and seven six models. um As in the historical warming patterns. We see the arctic amplification of warming um and now we're also seeing antarctic amplification. But we still see this hemispheric asymmetry of greater warming in the arctic than in the antarctic.

C

So why is this happening um to look into this I'll use a warming contribution analysis which has been used in these papers as well, and this is based on a local energy budget where we've calculated the co2 forcing um also have calculated different climate feedbacks and atmospheric heat, transport and ocean heat transport and uptake as well as a residual term, and the goal of this method is to try to express all these different factors as contributions to the total near surface forming in the arctic.

C

So we split up the the plank feedback uh into a global mean blank response and the local deviation from that global mean and then divide. Everything by the global mean plank feedback, and this gets to units of kelvin. um And if we rearrange, we can express the total near surface warming in the arctic um as coming from different contributions from feedbacks forcing an atmospheric and ocean heat transport as well as a residual term.

C

So we'll look at each of these different factors in this next plot um and here we're looking at how each factor contributes to tropical warming versus arctic warming. So there's this one to one line um and if a factor is farther away from that line in the upper left triangle, it's contributing more to arctic amplified warming.

C

So you can see the lapse rate and albedo feedbacks are contributing most to arctic amplification in these models in summit 5 and student 6, and this is consistent with previous studies um and just for a reminder on the lobster feedback. This is based on having these base state near surface temperature, inversions, so really stable. Temperature profiles that um inhibit vertical mixing and they trap um under co2. Forcing this. uh This warming is trapped near the surface.

C

So um as a result of the surface amplified warming there's really an effective long wave cooling to space um which enhances that surface amplified warming compared to the tropics, where there's greater warming aloft and effective long wave pulling to space.

C

um So that leads to this negative lobster feedback in the tropics compared to a positive lobster, feedback enlarged um one different thing here is that we're using radiative kernels, which are the uh the radiative sensitivity to a change in albedo, for example, or other factors for each of the feedbacks, but we're using um observationally based radiative kernels that are consistent with the observed mean state, and this gives a slightly larger albedo feedback in the arctic than in previous assessments and it's more on par with the lapse rate feedback for its contribution to arctic amplified warming.

C

Another thing we did here was to separate the atmospheric heat transport into a moist and dry component, and so we can see as a result of doing that that increased latent heat transport, this moist atmospheric heat transport, is the third largest contributor to arctic. Amplified warming in these models and one way to understand that increase in latent heat transport is through a diffusive perspective of flame energy.

C

um So this is looking at vertically integrated, specific humidity times the latent heat of vaporization for the pre-industrial control and abrupt four times co2 experiment and then also their difference on the bottom. And we can see that the tropics as a result of starting out warmer have a greater increase in latent energy than at the pools. And this gives this increased equator to pole, gradient and in latent energy that results from a diffusive perspective and increased moist atmospheric heat transport to the arctic under co2.

C

Forcing um if we switch perspectives to looking at arctic versus antarctic warming and these hemispheric gas imageries. We can see that most of these factors contribute to stronger warming in the arctic than in the antarctic and particularly the lobster feedback. So a lot of papers have looked at this asymmetry and we argue in hanadal 2020 that um the lobster feedback is weaker in the antarctic as a result of antarctic elevation.

C

So incoming atmospheric heat transport at 60 south is shown here and it peaks at um elevations below the the height of the antarctic ice sheet in many places, and we argue that as a result of this, um there are on average shallower and less intense inversions in the antarctic, that give you um a lower degree of surface trapped warming and a weaker lapse rate feedback in the antarctic compared to the arctic.

C

um But more details are in this paper and other papers for that. um Another interesting feature here is that the moist atmospheric heat transport is one of the only things that's actually contributing more to antarctic warming than arctic warming, and we can see that this is the largest contributor to antarctic warming here.

C

So one way to understand these differences between the antarctic and arctic is again looking at the gradient, the meridianal gradients and latent energy changes, and because the antarctic starts off colder and drier, and also has less of less warming under co2, forcing there's less of an increase in the latent energy and moisture there, and so there's a greater equator to pull gradient um to the antarctic than to the arctic, which could explain this um stronger increase and moist atmospheric heat transport to the antarctic compared to that arctic.

C

um So that's one possibility. It's also possible that having all these other uh more negative feedbacks like the cloud feed, the shortwave cloud feedback and the antarctic could contribute to greater moist atmospheric heat transport to the antarctic as a result of having this greater equator to pull gradient and radiation.

C

um So some interesting differences between the arctic and antarctic. Here I will just end this uh project by looking at the intermodal spread in an arctic warming, um but we found similar results uh looking at the antarctic as well for this, um so we're showing how each warming contribution um its variants and on the on the diagonal, as well as covariance, between warming contributions, um and this is looking across models.

C

So every term is also normalized by the total arctic warming variance um to give the fractional contribution to warming variants from each term, and we can see looking at the diagonal that the albedo feedback is the single largest driver of intermodal spread, but looking off the diagonal at these covariances, and we can also see that covariance between the albedo feedback and lobster feedback also contributes substantial entry model spread.

C

So, models that have a greater albedo feedback have more surface warming and tend to have a larger lobster feedback which amplifies inter-model spread.

C

If we look at covariance between the albedo feedback and dry heat transport that actually damps intermodal spread and dry heat transport has a negative covariance with almost all the other warming contributions. So this suggests that dry heat transport is kind of responding to arctic warming.

C

So if you have greater warming in a model, um there's more of a reduction in atmospheric dry heat transport to the pools which stamps that arctic warming and damps the intermodal spread- and I really like this figure- because it shows that these different mechanisms are are connected um and also shows that it can be helpful to use more idealized experiments to try to pick them apart and look at interactions between them, which I'll talk more about in the second um second paper. I'll talk about.

C

But first to summarize this um first section we looked at contributions to polar warming and hemispheric asymmetry and saw the lobster and albedo feedbacks are pretty key for greater warming in the arctic than the tropics or antarctic and also moisture. Transport um is an important contributor to antarctic warming, and then we just saw that intermodal spread is amplified by the albedo feedback and its covariance.

C

With elapsed, feedback and damped by dry, atmospheric heat transport, so next I'll move on to seasonalized symmetry and looking at what drives seasonal asymmetry in arctic warming, and this plot is showing near surface temperatures for different periods in this um cesm2 experiment, with increasing co2 concentrations at a rate of one percent per year, up to a quadrupling and then constant concentrations after that, and then comparing each of these periods to the pre-industrial control um gives these near surface temperature changes in the bottom plot, and this is showing some key features of the seasonal pattern of arctic warming, with greater warming in winter than in summer and specifically in early winter.

C

Having peak warming early on shifting to late winter um later. In these experiments to understand these patterns, some mechanisms that have been proposed are seasonal, ocean heat storage and ice insulation effects, as well as um long wave cloud and temperature feedbacks that peak in winter. But here I'll focus on this mechanism of having increasing effective heat capacity of the surface layer as you transition from ice to open ocean.

C

So um how how this mechanism works is that there's um the ocean mix layer has a large effect of heat capacity as a result of having this thick layer of water. That's coupled to surface heat, fluxes um and melting sea ice also has a large, effective heat capacity because all of the energy that um you add goes towards latent heating to melt ice rather than raising the surface temperature, um in contrast to open ocean or melting sea ice frozen sea ice has a small, effective heat capacity and acts like a shallow mixed layer.

C

So um the large, effective heat capacity of open ocean leads to this slow rate of seasonal warming and cooling all over frozen ice there's much faster rate of seasonal warming and cooling, and the idea is that, as you, transition from frozen ice to open ocean, um you slow this seasonal heating and cooling rate, um which gives you much warmer temperatures, particularly in early winter over ocean compared to frozen ice and gives you this peak early winter warming.

C

But this is just one of many different mechanisms that have been proposed so to try to pick these apart. We used a single column series model and also some idealized experiments and a gcm to test out this mechanism, so I'll focus just on the single colon model here um and comparing the single column model to cesm2 that we were just looking at, but here at 90 degrees north.

C

um We can look at these seasonal patterns of warming, so looking at temperature surface temperatures for different forcings and also um the anomalies for each forcing compared to the pre-industrial control or zero, forcing experiment, we can see that the single column model generally captures the seasonal pattern of having greater warming in winter than summer and peak warming and early shifting to late winter and in the single column model. You have frozen ice for temperatures less than zero and then melting ice for temperatures equal to zero and open ocean for temperatures greater than zero.

C

So this model has some um seasonally varying feedbacks, but even if we get rid of seasonality and feedbacks by using an annual mean plank feedback and keeping the albedo um equal to the albedo of ice, a constant ice albedo, um we can see that we still get the seasonality in warming even without seasonality and feedbacks and in part, that's coming from having uh increasing conductive heat flux through thinning ice in winter.

C

But even if we keep the um oops sorry, even if we keep the ice depth constant and the conductive heat flux in this model- and we have constant warming over frozen ice. Simply transitioning from ice to open ocean will give this pattern of peak warming in early winter um shifting to late winter. So this suggests that this is kind of a fundamental feature of having the surface layer shift from ice to ocean and having that increasing effect of the heat capacity, but to try to explicitly model those changes in effective heat capacity.

C

We used an even simpler version of this model where we had no cs and only an ocean mixed layer and constant ice albedo, and we just wanted to model these different. Effective heat capacities of the surface layer, so here we're again looking at surface temperatures and changes in surface temperature compared to a stereo, forcing experiment and we're representing the small, effective heat capacity of frozen ice with a shallow mix layer and then the larger, effective heat capacity of open ocean with a deeper mixed layer here.

C

So we can see this fast rate of seasonal warming and cooling for the small, effective heat capacity of frozen ice, represented versus a slower rate of seasonal heating and cooling for the larger, effective heat capacity of open ocean, and if we allow for a transition from um this small, effective capacity of present ice to the larger um thermal inertia of open ocean.

C

This leads to a slowing of the seasonal heating and cooling rate, and so you have these big differences between temperatures over open ocean and over frozen ice that gives this peak early winter warming in this model and later on in in late winter, as you shift from mice to ocean.

C

So the key point here is that simply increasing the effective heat capacity as the surface layer, transitions from ice to open ocean can produce this seasonal pattern of arctic warming. So that was our first key point.

C

We also ran some experiments that I'll just rush over here, um but we had some experiments in csm2 or sorry in csm1, with um standard experiments with ice and some experiments with no ice, but with albedo changes identical to the experiments with ice, to try to pick apart the role of sea thermodynamics, which generally suppresses warming in summer and increases warming in early winter.

C

But we saw that um through these experiments, the effective heat capacity also impacts seasonality and arctic warming by inhibiting a summer. A positive summary lobster feedback. So um there's also some lobster feedback seasonality coming into into play here.

C

um But just to add this to the summary, uh we saw that um increasing effective heat capacity of the surface layer can alone produce the seasonal pattern of arctic warming. Although there's also other factors contributing to seasonality in the arctic, yeah I'll just leave up the summary, but thanks for listening and happy to take questions.

D

Thank you so much lily. We have time for questions.

D

You have a question, feel free to either use the raise hand, button um or type your question in the chat or type. I have a question in the chat and we'll call on you. I'm trippy go ahead.

A

Yeah thanks for a great talk lily. um I was intrigued by the first study you presented, so um given the fact that the models seem to be overestimating warming in the antarctic compared to observations.

A

Do your results suggest that moisture transport is being overestimated by the models or how would you start to tackle that type of a question.

C

Yeah, I think that's a big question of like why the the model to um observation differences in the antarctic and there's like many different possibilities, including like internal variability um as part of that, so I think it might be less about the the moisture, the atmospheric heat transport and more about ocean um circulation uh there, that the models aren't capturing. But I think that's like a huge open question of like what exactly has been going on in in the past.

C

In order to better predict what will happen in the future in the in this other notion, and we didn't.

A

Really just go back.

C

A

C

Starting for paleo.

A

C

D

We have a question in the chat which I can read out loud um uh from alex thompson who says. Thank you for the nice talk. You showed that albedo was a much stronger feedback for the arctic than for the tropics. Do you think future changes in arctic vegetation, such as boreal forest, overtaking tundra, I.e, lowering albedo, will lead to an even even larger warming asymmetry between the arctic and the tropics.

C

Yeah cool, I think that's really interesting question and has been understudied. I think, compared to these other radiative feedbacks thinking about like land climate interactions, so I'm working on a project related to that right now, but um yeah there's some interesting uh findings already.

C

That um show that when you have when you include these like um physiological changes in the high latitude continents, that can lead to extra warming over those continents but can also lead to increased atmospheric heat transport to the arctic and greater warming in the arctic um as a result and then kicking in these other feedbacks like the albedo feedbacks but yeah, I think they this um park at all 2020 paper found that that's like uh like gives you like, 10 greater arctic warming than just the radiative um feedbacks, but that's more just looking at physiological like plant changes um in their pores and also leaf area index, and I don't think that's, including like the expansion of forests, um but I think, like abby swann has some papers on that affecting both albedo and evapotranspiration um and yeah.

C

I think that's a really interesting question. That's been less explored than these radiative feedbacks.

D

We have another minute or so left for more questions. Go ahead.

B

Thanks for the wonderful talk, I think the correlation between different feedback mechanisms is really interesting. Can you say something about maybe on the causality between these different processes? Thank you.

C

Sure yeah, so I think um nicole feldel and robin bakey have more work on this, um but at least specifically looking at like the albedo and lobster feedback, I think people think that the causality is sort of just the the sea. Ice changes are what's most important um and then, if you have greater sea ice loss, that gives you a stronger both a stronger albedo feedback um and also more as a result of that and just the base. The the sea ice loss itself.

C

Having greater uh near surface warming, gives you this this stronger lobster feedback so yeah. I think the like root cause of that all is the sea ice loss um and they've looked into into that a lot. So it's kind of like picking a part, that'll be done. Laptop feedback, maybe doesn't make so much sense, since they definitely interact and the lobster feedback is dependent on the albedo feedback um yeah. So that's one example, and then I think for the causality of the dry heat transport. I think it's the um like a response to arctic warming.

C

So if you have more arctic warming, you have less um less of a like equator to full gradient and um dry static energy, so you that reduces your dry heat transport to the to the arctic. As a result of that um yeah there's a lot of different things to look at here, though, with the different interactions so yeah, I think some more idealized experiments can be helpful too.

B

D

Great, um I think, that's about the time that we have for questions, so we can go ahead and move on to our next speaker. um John. Do you want to do a quick introduction.

B

Yeah, I have did the introduction. So next speaker is professor elizabeth. Please take it away. Thank you.

D

And eli, when, when you have two minutes left, I'm gonna hold up this post-it which might be backwards for you. But you can guess what.

E

Great, I hope you can hear me: okay, nice, so I'd like to talk about equitable climate, explain what the challenge is and suggest two ideas for for um explaining these challenges with equivalent climate in particular. This is based on mostly the work of dorian abbott when he was a student and and tim cronin when he was, I posted with me a few years ago.

E

So here's the main challenge. In the present day. um You look at any of the northern states of the united states and you have these events of arctic air that is coming from canada, very cold temperatures. This is the specific event in 2014, and this is from some weather um tv station, and you see that with windshield temperature is minus 40 to minus 65.

E

You know, fahrenheit celsius doesn't matter at this point. It takes a five minutes for your screen for you for your skin to freeze if you're outside okay. So that's the present day. This happens a few times every winter, at least very, very cold.

E

Here's what happens in equitable climates, so we're talking about you know 50 million years ago, was the peak of the eocene, and so let's look at this here. This first talks about the mean temperature. This is not the challenge, yet this is just background.

E

So what you see here is temperature of the deep ocean as function of time. This is the present. This is 60 million years ago. This is the eoc, and you see that the deep ocean temperature is estimated to have been about 12 degrees celsius and because the deep water temperature is determined at the high latitudes. It means that the high latitudes were very warm.

E

This is uh strengthened by this um picture right here, so there's a proxies um and mostly fossil evidence of things like crocodiles and palm trees and other plants and animals that cannot sustain sub-freezing temperatures. So you know if the temperature goes down below freezing for a night or two palm trees can survive and crocodiles and alligators, and so on can survive. The numbers are right here and define the fact that we find fossils. Of these, things vary at very high latitudes.

E

Whenever there's a black dot basically means that the temperature in these areas, such as in green river wyoming, which is a famous fossil site. The temperature in these areas never went below freezing for millions of years, and that is the challenge. How could it be that, even in the coldest day of the winter, for me for thousands and millions of years, the temperature never drops below freezing, and this is in the middle of the continent, far away from the moderating effect of the ocean.

E

So this is the challenge, we're trying to explain how could these things survive and how could the temperature not go below freezing even briefly for such a long time, so I'd like to propose to a two-part answer, part one based on these papers. Is that what happens? Is you have marine air coming from the ocean very moist in a warm climate, the ocean would be very warm. This marine air goes over the continent.

E

The moisture condenses creates low clouds, and these low clouds have a radiative effect that prevents the surface from cooling and by the time the air leaves the continent say it's coming from the pacific toward north america and then continuing to the atlantic with the westerlies.

E

Then, by the time it leaves the the continent it's replaced by other warm and moist air. That creates these clouds and these clouds are what's forming. So that's part, one of the answer that I'll try to explain part two has to do with the arctic. If the arctic is ice covered, then air coming from the arctic won't be moist, because it won't be able to absorb moisture from the ocean.

E

So we have to make sure that the arctic is ice free in an equitable climate and the second mechanism, I'll propose, is convective cloud feedback that creates a greenhouse effect during winter time that prevents the arctic from freezing in in an equivalent climate. So this is the two part answer again: low clouds keep the continent warm and and the convective clouds keep the arctic ice free and allow this moist air to form in order to later cover the continent.

E

Okay, so part, one. Is this arctic s, air suppression, by low cli cloud? So let me explain this is the work with team cronin. So let me explain this, so we start by taking a simple single column model worth. Basically, if someone cares- and we pretend that this represents an air column starting over the ocean and moving over the continent, the day is zero and it's moving over the continent where there's no shortwave radiation, because it's winter time. So this is the temperature profile of the model at t equals zero at day equals zero.

E

You see that it started zero the surface. This is called winter ocean and then, after two days, the temperature warm cools to this profile and you see that it cools at the surface by 15 degrees within the first day and within a few days, it's cooling to minus 40 degrees celsius.

E

So this is the process of arctic air formation. Now this air is then adjected from canada to the united states and and might have killed these dinosaurs and at the earlier phases, or later, the the palm trees and crocodiles and so on. So what happens in a warmer climate? So this is the same simulation same single coil model except this time, we're assuming that the ocean is a 20 degree we're taking an extreme to see what happens. So this is day zero.

E

This single column model, the surface is a 20 there's some lapse rate and then we need within the first day, this cools again by 10 degrees. So it's still a very significant cooling. But then, if you see day two day, four and so on, there's hardly any cooling going on and the reason there's no cooling in this case is that clouds are formed. The clouds are denoted by these green patches here. So we have lots of moisture in the air. That's coming from starting very warm over the ocean that moisture condenses.

E

These clouds have a strong radiative effect that I'll show in a second that prevents further cooling for a while by the time the moisture, dissipates and the cooling continues. The air is off the continent and replaced by other air. So that's the mechanism we're proposing this. This shows this a little bit better. Perhaps so, what's shown now in this case, is temperature of the surface as function of time.

E

You can see here 14 days, so you see this initial cooling that happens in in this warm ocean case, so an abrupt cooling in the first few hours, but then there's a plateau in which there's hardly any cooling and then later the the cooling continues. The reason there's no cooling here is, if you look at the net long wave radiation.

E

This is this orange curve. You see that the long wave radiation starts very strong, so there's very strong cooling at the beginning, but within hours that stops when the clouds form and then the clouds prevent any long-wave radiation, and that's why the surface doesn't cool for a long long time. Eventually the clouds dissipate and the cooling does continue.

E

This again shows the clouds form, so you can see the cloud forming within a few hours within the day and then we have sustained cloud for about eight days, and these are the clouds that keep the surface warm okay.

E

So this was a single column model so to to test that using a little bit more elaborated model. We take a three-dimensional model. Now we prescribe the sea surface temperature, so this is a series of four out of many runs that we've done beginning with pre-industrial temperature and going all the way to a prescribed sst where the temperature never drops below 20 degrees celsius. So that's an extreme high temperature scenario and then we calculate the temperature everywhere, in particular over the continent, where we're isolated from the warm warming of the ocean.

E

So here's what we find this is the minimum two meter temperature in the pre-industrial case. You can see these blue colors correspond to -40 degree celsius, and this is the case. When we have a warm ocean, you can see that the continent is much much warmer harley hunter drops below zero. You can also see the pdf, so this also tells you about the variability.

E

This was the pdf, the probability distribution function of surface temperature in the pre-industrial climate and the red one is the pdf in the warm climate with the warm ssd, and you can see the temperature hardly ever drops below zero. If you look at the reason for this, you can see that this is low cloud fraction developing over the continent, so the gcm is consistent with a single color model because it shows low clouds forming over the continent and these low clouds. This is now the temperature profile.

E

This is the pre-industrial with the inversion, the well-known inversion at high latitudes. That lily also talked about- and this is what happens at the warm sst where the inversion is completely illuminated, then the surface doesn't go below freezing. So all of this is consistent with the arctic air suppression, by low clouds that was described by this single color model. This time in a three-dimensional gcm all right.

E

This is also relevant to the polar amplification mentioned before so, as lily mentioned, the lapse rate is an important feedback where, if you go to high enough temperature in maybe in a different climate, you know rcp, 8.5 or equivalent climate in the past, we find that these low clouds can eliminate the inversion, so an important feedback for explaining the the arctic amplification and the lab. Sorry feedback is going to be surface clouds developing because of the advection of more marine air from the ocean.

E

Okay. So this summarizes this first part. Basically we're saying here that um we find that um arctic air formation is suppressed. If the sea surface temperature is above 10 degrees celsius, as you expect it to be in a warm climate. So if the ocean warms in an equal climate, you can also explain the warming of the continent and, more importantly, you can explain why the temperature never goes below freezing thanks to the warming effect of these continental clouds, slope clouds that develop- and this might be relevant to future climate as well.

E

Okay, so stepping back and getting back to the original picture, remember we talked about um air coming into saint north america, and if the air is coming from, say the pacific in this direction and the pacific is warm, then that's fine. The mechanism I just described works because the ocean will be warm and and and the air would be moist and that would allow these low clouds to develop.

E

But what happens if the air comes across from siberia across the arctic as it does in this event, that was analyzed by all in that case, if that happens, in an equitable climate and the arctic is ice covered, then the air won't be able to accumulate moisture and the mechanism I described before won't work. So we must eliminate sea ice in the arctic during winter for this uh idea to work. So let me explain how we can get rid of sea ice in an equivalent climate, and this is the second part.

E

This is the convective cloud feedback mechanism. That's the work with dorian out okay. So we start with this. You write a simple energy balance for the arctic. You take into account. um You know they hit like hit from mid-latitudes it tracks for mid latitudes, they're different radiative fluxes, and you ask okay.

E

What do I need to do in order to increase the temperature of the arctic by 15 degrees, as happened in equal climate, and you try to change the heat transport from mid latitudes, the emissivity of clouds, the co2, the co2 with water vapor effect, and you find that the only thing that can do it essentially at the reasonable cost um that the using a mechanism that can believe is changes in the cloud fraction over the arctic during winter.

E

So the back of the envelope suggests, and the calculations suggest that clouds over the arctic ocean would be the most effective way to warm the arctic by 15 degrees and keep it ice free during winter. So, let's see how we can do that, so just a quick reminder about height clouds and low clouds. You remember that high clouds have a low albedo but high emissivity, so they have a warming effect on climate, while low clouds have albedo and have a cooling effect on clouds. So what we need is high clouds during winter.

E

Over the arctic to to study that we formulated basically a simple six box model we'll get to gcms in a second, but we started with a six box atmospheric model and we analyzed this model in the following way. So we plot here in the box model the temperature difference between the arctic and the tropics.

E

That's this axis right here, so again: temperature difference between the arctic and the tropics as function of co2. As we increase the co2. We start with the high value of this temperature difference. A high value of this temperature difference means a climate in which the temperature in the arctic is called so high values here correspond to cold arctic climate. Now we start increasing the co2.

E

You see the temperature difference between the arctic and the tropics is decreasing, but eventually there's a jump and then, as you decrease the co2 you get this hysteresis and at some value of steel, two you have two different states. One state has a high temperature difference between the arctic and the equator. This is a present day like climate and another state has a low temperature difference between the arctic and the low latitudes, and this would be an equitable climate solution.

E

When we look into this very, very simple box model to see what is the reason for these two states, we find that the equivalence solution with a small equator default temperature difference has clouds and convection above the arctic.

E

So this is a very, very simple model, but it predicts that in an equal climate, there would be con convection during wintertime over the arctic and that convection would lead to a very warm arctic. That's a obviously a dramatic prediction of wintertime convection. That is like the convection that happens over the tropics today, but happens during winter with no sunlight in the arctic. So it's a bit crazy and when we submitted this to um the publication, here's some reviews we get first controls.

E

The first reviewer tells us that this behavior will probably not be of much scientific interest. The second one said that the model simulation is incompatible with observations. The third one said that the model- it's not clear, why this simple toy model would get such a state while a realistic gcm cannot and another. One said that the basic the answer is built into the model and basic data about cloud properties have been ignored. So that was the enthusiastic reception we got in the reviews.

E

Obviously I wouldn't show you that if we were wrong, I think we're right, and these reviews were wrong and I'd like to show you why and for that purpose. Let's look at some ice. Icp ipcc runs. um So, let's see all right, we have time. So these are two models run in a four-time steel, two scenario: the end car model in the gfdl model and we're seeing we're looking at the arctic during winter time, and let's go one by one and see what the models predict the first model.

E

The encore model predicts that at four times co2 the arctic is ice. Free, that's what the red stuff here means. Well, the gfdl predicts that the arctic is ice covered. You continue and you look at the surface temperature. The anchor model shows the warming at four times co2 of about 25 to 30 degrees celsius. Why the gfdl model doesn't show any warming or very weak warming. The next one is the cloud radiative forcing so the arc.

E

The anchor model shows significant cloud radiative, forcing of 20 to 25 watts per meter square that keeps the surface warm and melts the ice, and the reason for that, as you can see in the last one- is convective precipitation.

E

So basically, one model shows convection over the arctic leading cloud to cloud radiative, forcing leading to warming and leading to ci smelting exactly as the box model predicted, and the second model shows the second state with no convection no cloud ready, forcing no warming and full sea ice, and if you need further proof that this convection will actually happen in the future, you can look at these from these results from a simi 5 model from erasmus work that we've been doing. So what you see here is five different models for each model.

E

You see as function of time going from present to 2300 and latitude. The colors indicate convective precipitation and you can see that every single model predicts that in the arctic at 85 degrees, north and 90 north, a convective precipitation will develop. Except for this one model I think that's the french climate module any other model predicts that in a warm climate in rcp, 8.5 convection will indeed develop over the arctic. According to this toy model that I just discussed.

E

Great and so I'll skip the summary of this one, because I want to provide sort of a big picture perspective into the whole thing. um So that would be this thing right here. Six ideas were suggested for equable climate so far, and I'd like to review each of them in 15 seconds and tell you what I think about them so that you have the big picture in mind. So this was the idea by kerry emmanuel that, in a warm climate, squirrel guns will get stronger.

E

This will lead to ocean mixing. The ocean mixing will strengthen the overturning circulation, and this will transport heat into the arctic. It's a nice idea that he suggested using a box model, but when he tried it with a gcn, the warming you get is about half a degree celsius, so it's completely negligible. So this doesn't seem to work. The next idea, the first one chronologically, is by brian farrell 1990 in a beautiful paper in which she suggested that if you have an equator to pull headless cell, this would lead to a very warm climate.

E

We tried that with the gstm. The thing is, if you have an equator to pull heavily cell, the subsidence over the pole leads to cloud-free skies. The cloud-free skies leads to significant warming and then sea ice develops and prevents this warming from happening. So this works in the dry model, but not when you take into account moisture and clouds.

E

Next idea was polo stratospheric class that was suggested by sloane, attal 92 and then further development by kirk, davidoff and alex 2002 carey manuel looked into this with a with a his student quality and they found that that particular mechanism doesn't seem to work for complicated reason. I don't want to get into that. So again. This one is an exotic mechanism. It's not clear that it's working!

E

Finally, um there's this recent idea by schneider 2019 that in high co2, stratocumulus, decks will break down and that would lead to a 10 degree warming of the subtropics. The problem with that- and I mean, I think, that's a beautiful work. Actually, the problem with that is what you need to explain is winter time, continental climate at high latitudes. Well, this deals with the subtropics and it's not clear how this warming will get from the subtropics to the high latitudes.

E

So to get back to the two things we talked about today, the first idea was the low clouds that suppress cooling and arctic air formation over the continents that I talked about and what enables that is an ice free arctic which is enabled by this deep convection during winter and over the arctic that we see in every five or six model.

E

That's it. Thank you.

D

Thank you. I was just about to hold up the two-minute sign, but you are ahead of time. Thank you for that excellent talk. uh We have plenty of time for questions, feel free to type them in the chat or raise your hand round. Go ahead.

F

Yeah um thanks for the super inspiring interesting talk. um Can you guys hear me actually.

E

Okay, yes, okay,.

F

um I was wondering it's a little bit content curative when you're talking about um uh transition to a warm climate, and you would start having even winter time deep convection in the arctic, because the current like seeing the previous talk like the global warming, actually increases the uh stability uh in the arctic.

F

So how does that link link to this development of voiced, convection.

E

Yeah, let me explain that.

E

Yeah, it's actually pretty pretty basic thermodynamics. So suppose you have two temperature profiles, and um so um suppose you have this temperature collapsed right and you just warm it with a constant shift of temperature, so no lab straight feedback, you're just taking the lab street and warming it as a result of that the moisture will increase much more near the surface, simply because of clausius glyceron.

E

You know, there's the same stream. Warming at the surface would lead to much larger increase in moisture as a result of that, the moist static energy will increase much more with the surface than it up in the upper atmosphere and that increase in in the moist static energy at the surface leads to a state that is convectively unstable and that would lead to to destabilization. In that sense, and you know you can see that in every you know it's not just this theoretical idea.

E

You can see it in every single um semi model that is being run, so all of them are showing that convection during winter time in the arctic. I do agree with you that completely that it's non-intuitive, but it's now been shown to happen in every model and the basic thermodynamics behind this is pretty simple. Actually, thanks for the question.

F

Justine, um can I ask you a quick follow-up question and you did mention that there is a hysteresis, so um it seems like under present condition. You wouldn't be entering this regime and but if you once you enter in this regime, it's hard to get out um what what what kind of process are at work.

E

The hysteresis is because of a positive feedback that happens because of surface warming that leads to uh um here. There's you warm the surface.

E

Here we go so you warm the surface. This makes the air column unstable. This leads to convection the clouds and further warming when you have a positive feedback like that, it allows for multiple equilibria to happen. I would say, though, that these there is a multiple equilibrium might be a the less realistic part of this. The important part of the mechanism is the fact that there is one state with no convection and with sea ice in another state, with a convection and nauseas, whether they coexist at the same parameter regime is less obvious.

E

It does happen in many simple models we looked at, but it's not obvious that it exists in gcms and it's not essential for the explanation of equitable climate.

F

Okay, thank you.

D

B

Go ahead thanks for the great talk you mentioned, the two mechanisms both involve this cloud formation, so my question is which one you think is maybe better captured by gcn and what's the if there's problem with gcm, it's more like a structural problem or resolution or parameter problem right. Thank you.

E

Yeah, so it seems the gcm do gcms do get recent gcms do get this convection in the winter over the arctic, so they do seem to be able to get that, even though they parameterize convection and so on. When we run these models with with very high ssd, they also get the low clouds over the surface. So it seems, like you know, both mechanisms are actually captured by gcns. Now you could ask how reliable is. Does this happen?

E

How reliable is the sco2 at which this happens, and so on that we don't know and when we do have great doubts about the ability of gcms to stimulate convection and cloud and so on. So I would say that, even though gcms do get both of these mechanisms, there are many uncertainties in terms of quantitatively at what co2 would these two feedbacks be triggered.

B

Thank you thanks.

D

A

Go ahead yeah. This is actually a follow-up on john's question. So um to get these feedbacks, it looked like you used, prescribed ssts, and does that point to an issue with the ocean component of the model in terms of getting the appropriate sst fields for these types of feedbacks to take place.

E

That's great yeah, you know so the way we did it is. First, we did a simple color model. That has nothing. Basically, you know just a single earth column that we pretend is moving from the ocean to the left. The next step is, we prescribe the sst and check the effect over the continent. The next step after that would have to be running a coupled model, increasing the co2, letting the model increase the sst and seeing how that affects the low clouds over the continent.

E

We haven't done that, but I completely agree with you that that would be the next interesting step to take, and I think there is going to be a challenge to get the ocean to warm enough. For this to happen, we find that you need to warm the ocean by at least 10 degrees with enough co2. You might be able to get it, but you know it's not obvious that you can.

A

Yeah, I guess a thought I had was like what extent of warming do you need? Because if you can get these feedbacks by just warming narrow coastal strips along the ocean, maybe resolution is key, but something.

E

A

E

Yeah, so the answer is here so this is this is the initial temperature of the ocean. So this is the initial temperature of the air over the ocean before it moves over land in this one-dimensional color model and, what's shown on the vertical axis, is how long does it take for this air to get to the freezing point? Okay? So if you start at zero, obviously it takes zero days to get to freezing, because you started that zero frequency. If you start with an ocean that is even nine degrees.

E

The initial cooling that is very that is very rapid before the clouds develop bring the temperature to freezing within hours. If the temperature of the ocean is above 10 degrees celsius, then the time to freezing becomes much larger. So, for example, if the ocean is 15 degrees celsius, it's going to take about a week to get this surf surface to be below freezing.

E

So you need to warm the ocean. According to this simple model estimate by by at least 10 or 12 degrees celsius, one would have to check this with the gcm. This is based on the color model.

E

So can you get the ocean to warm by 12 degrees celsius by increasing the co2 resistant to reasonable values? I guess it depends what we call reasonable right. It's not obvious.

A

Yeah, thank you.

D

I have a question maybe related to that. um I'm wondering if any of these feedbacks that you're talking about would potentially be further enhanced by the paleogeography in eurasia um and if that's something that you it looked like, you were using modern continental configuration, but maybe that's an additional influence.

E

Yeah, you know you wonder whether the the elevation in wyoming this green river green river fossil site, where they find all of these evidence and others.

E

I don't actually know what the paleo elevation of that is, that could have made a big difference. Obviously, so I think when you try to interpret proxy evidence, the topography is going to make a big difference. I think the idea of frost, intolerant plants and animals in high latitudes is pretty robust, regardless of elevation changes.

E

So I'm hoping that's not necessarily a huge issue.

D

Right now I was thinking more about in europe and asia. There's um because there's a few kind of inland seas and just the arrangement looks a little bit different there, and I wonder if that could have influenced moist.

E

D

To the arctic and therefore maybe influence some of these feedbacks.

E

That's right: they might, there might have been in a big inland sea in north america. Obviously, in the western part of north america that could have helped, because then you could accumulate moisture over land. You don't have to accumulate it over the ocean in order to form these clouds. So I think it does make sense to repeat such calculations using a gcm that uses the the paleo topography as opposed to the present-day topography which we've used. That was a bit crude.

D

You have time for one last question: john dionica.

B

I have a question for lily. I think eli's mechanism largely focused on the cloud feedback and there's this very warm climate state, and you talked about multiple mechanisms that may contribute to the polar amplification.

B

I'm wondering do we have any speculation about the air potential change under this extreme warm background climate, for example, including this moisture heat transfer. You mentioned.

C

Yeah good question um yeah, I know there's some people have explored changes in the albedo feedback under um like in the future and non-linearities with that, where, um as you as you warm more and start to lose more sea ice later on um in the winter and early spring, the sea ice loss aligns more with the like timing of like peak insulation, so that you have a greater albedo feedback um and, although, like your albedo, feedback is declining over time when you have, as you have less um sea ice to lose, um that sort of gives you like a a peak bump in the albedo feedback um and polar amplification once you shift to uh later or warming in later winter, early spring, um but yeah.

C

I would expect these other feedbacks to become muted over time when you, in these, like warmer climates without any sea ice loss. So yeah and it sounds like super non-linear for the cloud feedback of where in present day and um near future warming, it seems less important than other feedbacks like the laptop feedback. But once you get these convective, um this convection it's more important um but yeah. I think there's some also. Some, uh like overlap with current um or more like near future, lobster feedbacks being dependent on the the clouds in the arctic.

C

So I think there's a paper by ib10 introduced your altmo that looks at um if you change this, the amount of super cool liquid and clouds in the arctic. How that impacts the lobster feedback um so which is something that eli, I think, was talking about a little bit um in equitable climates as well yeah, that's something to think more about! It seems like there are a lot of nonlinear changes.

B

D

I think we are out of time. Thank you, everybody for the great discussion and thank you so much to our two speakers for really excellent talks, um stay tuned for our next webinar, uh which we will be announcing sometime soon. If you have speakers that you would like to nominate, please do so we're happy to take nominations. You are welcome to self-nominate as well.

E

Thank you. It's.

D

Good to see you.