 So this talk is on the challenge of modeling liquid bearing clouds in the Arctic and their implications for decadal climate variability. And I'm going to be focusing on the Arctic amplification. That's defined as the increase in the Arctic temperatures relative to the global mean. And this has been recognized as an important phenomenon since 1896. And if you look on the right here, this shows the surface air temperature anomalies from the last decade, whoops, minus the 1960 to 2009 mean, the upper figure shows the wintertime average, December, January, February, and the bottom shows the summertime average, June, July, and August. And the insets here show the zonal averages. And as you can see, we're already seeing this amplification in the Arctic over the last decade. And it's largest in winter as opposed to summer. And that's really interesting because that means that the sea ice albedo feedback cannot be responsible for the Arctic amplification. The sea ice albedo feedback is when you have a retreat in the sea ice and you expose the ocean, you could have increase in the incoming shortwave radiation, which will heat the ocean and cause the sea ice to retreat even farther. So that can't be responsible because there's no incoming shortwave radiation in the winter. Another really interesting result is that there's a very large spread in climate models of the magnitude of the Arctic amplification. And that's because these processes, the sea ice albedo feedback, and all of the other processes that must be operating are represented very differently in all of the climate models. So there's a large spread in the projected magnitude of Arctic amplification. Another interesting result is from the K et al 2015 paper, which looked at the CESM large ensemble, which was discussed yesterday, and found that the spread of the Arctic amplification across their ensemble was so large that over 50% of the observed trend can just be due to natural variability. And that's really interesting because, as was commented on yesterday, this is a micro-ensemble. So there's just perturbations in the atmospheric conditions. And it'd be very interesting to sample the macro example ensemble like from the previous talk that samples different conditions of the AMOC and the ocean conditions. So I'm going to talk about some of these other processes that must be operating in order to create the Arctic amplification. I'll talk about seasonal redistribution of heat and the role of surface inversions. And then I would like to focus on the role of the liquid-bearing clouds because liquid-bearing clouds are ubiquitous in the Arctic, but they're incredibly difficult to model. And it's really interesting. It wasn't even observed until 1997, 1998, in the surface heat budget of the Arctic experiment, Shiba, that you could get super cold liquid in these clouds at temperatures down to minus 40 degrees. And it was thought before that that this wasn't even possible that you could have ice and liquid at the same time at such cold temperatures because the saturation vapor pressure of ice is lower than liquid. So the ice would remove all the vapor from the atmosphere and not allow liquid to form. But that's actually not what happens in reality. You get these liquid-bearing clouds all the time in the Arctic at very cold temperatures. And then I want to show with a Greenland example how important it is to really look at cloud impacts beyond just cloud-radiative fluxes, that they have many other impacts that can affect Arctic amplification. And if there's time, I'll discuss whether this might be an interesting challenge for CliveR. So this shows the seasonal redistribution of heat. On the left, this shows the energy budget at the top of the atmosphere. And on the right, it shows the surface temperature. And this is the Arctic mean from 13th CMF3 A1B minus 20th century simulation. So this is the response to external forcing. And if you look on the left first, the dark shading shows the long wave fluxes at the top of the atmosphere, and the light shading shows the short wave. And if you just focus on the shaded boxes, the leftmost box shows the annual mean, which you get this balance between long wave and short wave. But if you look at the red box here, which is the summertime mean, you can see that there is this increased downward short wave radiation in summertime. But this is happening when the surface temperature anomalies are smallest. And you're getting, in wintertime, you're getting this outgoing long wave radiation from the surface. And that's happening when the surface temperature anomalies are actually largest. So you're getting incoming short wave radiation in summer, which is then radiated to space in the winter time. And this is from the same paper. It shows, again, this is the zonal average now. And so the dark line shows the changes in the surface air temperature. And the gray line shows the changes in the ocean mix layer. And what this shows, you can see this very large arctic amplification in the surface air temperature. But you're not getting an amplification in the ocean mix layer. So this is showing that the ocean is not warming up. And that's not responsible for the increased warming at the surface. And so there must be other processes at play here. And one of these really important processes is the role of surface inversions. And this shows the sensitivity of arctic amplification to inversion strength. And again, this is a doubling of CO2 experiment. And again, on the left here, this shows the zonal averages. And so what this study looked at is that you have an increase in ocean heating in the summertime, which is then released through the ice as it gets thinner into the atmosphere and then is trapped at the surface because of the surface inversion. And then the arctic, especially in winter time, you're getting surface inversions all the time because you get this very efficient radiation to space. And you get much colder temperatures at the surface than aloft. Also, you get advection of warm, moist air masses above the cold surface. And so these processes cause these very strong inversions. And what this study looked at is if you have an increase in turbulence at the surface and you can increase or reduce, or reduce the inversion strength, what impact that has on the arctic amplification. And so if you look at this zonal averages, the black line here is the control run. So you're getting a warming in the arctic of about eight degrees, seven degrees. And when they decrease the mixing and increase the inversion strength, the arctic amplification increased to about 10 degrees. And if you increase the mixing and reduce the inversion strength, that decreased the temperature in the arctic to about five degrees. So this is just showing the sensitivity of arctic amplification to the inversion strength. And that's what's illustrated here on the right. So this shows the changes in the top of the atmosphere radiation. The left is for increased mixing. The center is for the control and the right is for decreased mixing. And if you just focus on the red bars, this is just showing that as you increase the mixing at the surface, you're mixing that warm air up a little higher where it's sufficiently radiated to space. So what is the role of liquid bearing clouds? This is from a climate model simulation. Again, a doubling of CO2. So the response to external forcing, increase in greenhouse gases from just one model. It's the GFD LCM 2.0, which had a pretty good simulation of these liquid bearing clouds. And what I'm showing here, these are, so this is month and this is latitude. So 45 north to 90 north. So this is the North Pole. And this is the change in surface temperature. The total change in surface temperature as a response to an increase in greenhouse gases. This is the change in the surface temperature due to the sea ice albedo feedback. This is the change in the surface temperature due to the cloud forcing. The change due to the clear sky net long wave flux. The change due to heat storage, which is the negative of the warming of the upper ocean. So negative here in summer means warming. And then the positive here means it's released to the atmosphere in winter. And then in the bottom right here, this is the change in turbulent fluxes. So first looking at the sea ice albedo feedback. In this model, at least, you have the sea ice albedo feedback operating in summertime, but you're getting also increased cloud cover in summer, which almost completely compensates for the sea ice albedo feedback. So even though this is a very important feedback, its impact on the Arctic amplification depends on how much the model simulate increased cloud cover as well. And it's really interesting these low level stratocumulus liquid bearing clouds in the Arctic. There's about a month where they actually have a negative feedback, where they actually reflect more short wave than they increase the downward long wave. But for the rest of the year, they actually have a positive feedback, a warming of the surface. So if you average over the whole annual cycle, these clouds actually warm the surface. So they play a really different role than these type of clouds in the rest of the planet. And if you look at the relationship between the total surface temperature and the cloud forcing, you can see in winter that about 40% of the total warming is due to increase in cloud cover because of the increased warming. And if you look at the relationship between the heat storage and the turbulent fluxes, as I showed there is a warming of the mixed layer in the summer and then this heat is released to the atmosphere in winter. But what this model is showing is that this increase in the surface air temperature is causing increase in turbulent fluxes and is not responsible for the increase in the surface warming. So it's what I showed in the previous slide that you're getting this mixing of the air to the atmosphere where it's efficiently radiated to space. And if you look at the net clear sky long wave flux, you can see that it's actually the largest term in winter. And this is really interesting because that means that the atmospheric temperature and moisture has changed and is making a larger contribution to the Arctic amplification than the cloud forcing, the CIS albedo feedback and the heat storage in the ocean. And this might be because of changes in transport into the Arctic or it could be changes in the atmospheric temperature and moisture because of cloud cover, mixing by cloud cover. So in observations, you see this bimodal distribution. If you look at these bivariate PDFs where you have surface net long wave flux on the left and the low level stability on the bottom. So low level stability is defined as the 850 millibar temperature minus the surface temperature. So it's an estimate of the strength of the inversion. And you can see these bivariate PDFs. So we have these two modes. This is from the arms site at Barrow. So this is a very long record that we have along the coast. And this is from the Shiba year, the surface heat budget of the Arctic experiment that I mentioned before, it's one year. And you can see that you get these bimodal distributions. You get a cloudy mode and you get a clear mode. And so the clear mode is because you're efficiently radiating to space and causing large inversions and large outgoing long wave fluxes. And you're getting a cloudy mode because these clouds are very, very low. They're about one kilometer from the surface. And the downwelling long wave radiation essentially compensates for the outgoing surface long wave radiation. And because you get cloud top cooling and very efficient mixing driven by the cloud, you're actually homogenizing the whole surface layer and removing the inversion. And so you get this very low level stability. So in observations, you see these very distinct two modes. And so this is a paper showing whether to what extent the models were able to represent these two modes. It's an indication of whether the models are able to represent these liquid bearing clouds. And as you can see, there are a couple of models that do have the cloudy mode here. But in general, that cloudy mode is completely absent in these, this is the CMIT 5, 11 CMIT 5 wintertime averages, 1990 to 1999, average over the ocean. So it's the closest thing we have to really estimating how well the models are representing these liquid bearing clouds. And you can see that this cloudy mode is essentially absent. So I want to give an example about how it's really important to look at the cloud impacts beyond just cloud radiative fluxes. And so this is an example from Greenland. This was the extreme melt events in July of 2012. And this is from the Benartz et al 2013 paper which showed that this melting would not have occurred if it wasn't for the existence of liquid bearing clouds. And so that's what's shown on the left here. The purple shows where the liquid bearing clouds were observed. And so you had melting at this time all the way up to the top of the ice sheet at summit which is this black dot here. And the last time that occurred was over 100 years ago. So this was a very extreme event. And on the right here this blue line shows the height above which melting would not have occurred without the existence of these liquid clouds. So we did a regional climate model simulation of this event to try to look at some of these additional impacts beyond cloud radiative fluxes. And I'm just showing here on the left the radiative fluxes from the observations. We have a very long record of observations from the summit station. This is just showing that July 10 to July 13. This is just the extreme event period. And on the right here I'm showing all the terms in the surface energy budget from the model. And this is just to show that the model is doing a really good job of reproducing the radiative fluxes that were observed at that time. And it's really interesting if you look at the purple line here this is the energy flux due to phase change. Just to show how extreme event this was. So about 60 watts per meter squared went into melting at the peak of this event. So stratocumulus over the ice sheet are, there's a liquid at the top and there's ice precipitating out the base. And the liquid at the top is causing this long wave cooling which drives the the turbulent mixing. And there's entrainment of air from the top of the cloud into the cloud layer, which could either be a moisture source or not. Which is a really interesting topic. But in order to remove the clouds what we did is we didn't allow the radiation to see the cloud liquid in the cloud ice just in the region of summit. We didn't wanna change the air mass properties at all as they're vetted to summit. We wanted to just look at the local impact of the stratocumulus at summit. And this is really interesting because it's always been hypothesized that these clouds can maintain themselves through cloud top cooling, through the turbulence driven by the cloud top cooling. But it's really never been proven but that's exactly what we found. So this is showing the cloud water content over this three day period from July 10 to July 13. And so you can see that you're getting these stratocumulus forming about a kilometer above the surface when you allow the radiation code to see the cloud water but when you don't allow the radiation code to see the cloud water, the radiation, the stratocumulus completely disappear and you do get condensation at night but the radiation code doesn't see that. So this is just to show the impact of the clouds on the temperature structure. So this is with the clouds minus without. And you can see that the cloud top cooling is causing a cooling of about five degrees. And the warming at the surface is warming it by about five degrees. So you're getting from the top of the cloud to the surface is an increase of about 10 degrees. And this is just the homogenization of the boundary layer which is completely eliminating the inversion. And as I showed before, it has a huge impact on Arctic amplification. And on the right here, I'm showing the sensible heat flux. The red is with the cloud and the black is without. And this is just to show that without the cloud, you're getting a cooling of the surface during the day and a warming at night and the average is about minus two watts per meter squared. But when you have this constant warming by this cloud cover, it increases the magnitude of the sensible heating to about minus eight watts per meter squared. So it's a six watt per meter squared increase just due to the turbulent fluxes, which as I showed before, has also a very large impact on Arctic amplification. And this just shows the impact on the ice temperatures. So if you just focus on the black line, that's the five centimeter temperature. And without the cloud, you can see that you don't get a net warming. So you're warming as much in the day as you're cooling during the night. So there's no net warming. But if you look at with the cloud, you see with the black line, this is constant warming. Because that's because the ice is integrating the effect of the clouds. And this is really important for the sea ice because it's reducing the temperature difference between the ocean temperature and the ice temperature, which reduces the amount of heating of the surface air temperature. So I've been focusing this talk on local processes, but there are just as many issues related to extra polar interactions, such as what the previous talk talked about, changes in ocean currents, changes in atmospheric transports. And a lot of the issues I've been talking about are partly related to the fact that the data is so sparse. And it's very difficult to constrain these processes with measurements. And in order to improve our understanding of Arctic amplification, I think it's really necessary to focus on these fast processes. Because these are the processes that are integrated and are responsible for determining the magnitude of the climate, the model response. And this might be a nice focused idealized model study to be able to constrain some of these processes in the future. Thank you.