MEGHAN F. CRONIN: Yes. OK. And I always really like to give live talks just because it's so much easier to read the room, and part of giving a talk is getting feedback on the work. And it's really nice to kind of see people squint their eyes and say, "hmm," "um." Anyway, I can't do that right now. So please, chime up if you have any questions or put them into the chat box. And I really look forward to talking with you about this. I'm going to be talking about the diurnal cycle of sea surface temperature in frontal regions. And I'll be bringing in some inner comparisons with the ERA5 product that Bo Yang has done, so I definitely want to acknowledge him. I think he's online. And my collaborators, Dongxiao Zhang and postdoc Samantha Wills and Jack Reeves Eyre from UW CICOES and LuAnne Thompson from UW. I also want to acknowledge Laura Braby, who is also working on some ERA5 inner comparisons on Agulhas Return Current mooring, south of South Africa. And then for the OASIS, that's the Observing Air-Sea Interactions Strategy, which is now a UN Decade program, I want to acknowledge Christa Marandino, my co-chair; Seb Swart is also a co-chair; and the full SCOR Working Group number 162. And Phil Browne is a member of that working group, so thanks, Phil. And let's get started. So the Earth's rotation causes a daily cycle in solar radiation, which acts as an external clock for the Earth system. And so what that means is that the surface of the ocean-- this is a very complicated place, because there's the skin temperature. That's what the atmosphere feels. But there's also, as you can see in this panel down below, there's a lot of stratification. There's first a cool skin, right in the skin, the microlayer. And then below that, in the daytime, under light winds, there can be a warm layer stratification. So that's like when you're in a lake and your chin is warmer than your toes. And then at the bottom is the foundation temperature. That's the sea surface temperature below this daytime stratification. So this is a fuzzy figure, but this is from a paper by Carol Anne Clayson and Bogdanoff showing that, if you neglect that warm layer, you can have large errors in your net surface heat fluxes. Even averaged over a full year, you can get like a 10-watts-per-meter-squared error in the Western Pacific Warm Pool. I mean, when we had TOGA COARE, that was the uncertainty that we were trying to get towards. And here we are, blowing it away just on the diurnal cycle. And then recently, Watson et al. showed that including the cool skin effect in the carbon fluxes can lead to large errors and, in fact, a 30% difference in the global uptake of carbon dioxide just by including the mean cool skin effect. And when he compared that corrected global flux to the uncorrected, it matched much better the inventory change of carbon dioxide globally. And so that just really goes to show the importance of getting this skin temperature right. And then now, Bo Yang, who's on the line here, has a paper that's under review that's showing the effect of skin temperature on oxygen fluxes. And that's because the skin temperature can impact the mass transfer coefficient for air-sea gas diffusion and the saturated oxygen concentration. And this can cause a-- well, it varies. He's looking at it from monthly fluxes, the differences with the BGC ARGO floats that he has, profiling floats. And in the tropics, subtropics, and the higher latitudes near Papa, it can have between a 1% and 30% difference in the flux. And that actually also leads to a difference in the annual net community production. So the way he's going about including this skin temperature is by extrapolating to the surface, extrapolating the temperature, the foundation temperature measured by the profiling float to the surface using the ERA5 delta difference, the skin minus the foundation temperature. And so the first question that he asked is, how good is the ERA5 delta temperature? And he used the Papa and KEO and a tropical mooring at 0 on the equator at 95 west during TAO-EPIC. All of these three moorings have flux capability, so they were measuring the bulk net surface heat flux. And as part of the COARE algorithm, we compute the skin temperature by doing a warm layer correction and a cool skin correction. So he was able to compare that skin minus sea surface temperature to the ERA5. And as you can see here, it's pretty good. The top is the Ocean Station Papa in comparison, and then there's KEO and the subtropics and 0, the Eastern equatorial Pacific TAO-EPIC mooring. Pretty good, but, but, but, but I will be showing-- there's a big "but" on this, and I'm not going to give it away until we get there. Oh, OK. Now, where are we? What to do? Let's look at the diurnal cycle of sea surface temperature. We have all of these moorings, not just KEO and Papa, but we have the whole array of tropical Pacific moorings. And you can go to this website and download the time series to get the high resolution. I'm here grabbing sea surface temperature. And let's look at some diurnal cycles. We've actually done this for all the sites. And these are plots that Dongxiao Zhang, who I think I saw that he was on the line. I'm not sure. He is calculating the seasonal diurnal cycle. So DJF, March, April, May. June, July, August. September, October, November is blue. In the 0, 165 East, the Western Pacific, Central Pacific, and the Eastern Pacific. And of course, the time zone changes. And so we're showing here the solar radiation. So you can see noontime here. Eastern Pacific has a really large diurnal cycle variation over the season. That's interesting. We'll be talking about that. OK. What else do we have? We have wind speeds. We can do the diurnal cycle of wind speeds. And look at that. This is really interesting, that, at 0 North, 95 West, there's a really large diurnal cycle in winds in SON, this September, October, November. Not so much at the other ones. What's going on? So this is another way that Dongxiao has shown the diurnal cycle. He's showing the seasonal cycle of the sea surface temperature. That's in red. And so you can see the red shows it's warm in the west and colder in the east, Eastern Pacific. And this is showing-- let's see. Oh, we have 8 North at the top, 2 North, and 0 North. So this is kind of like a map. The lower right is the-- oh, 0 North, 95 West. These surface temperatures seasonal cycle, and there's a very big seasonal cycle. And then on top of this is the diurnal cycle range, shown in black. And so in September, October, November, that's when we have the cold tongue is fully in force whereas the cold tongue recedes in January, February, March. That's when we had the highest diurnal cycle, but it's still really large in September, October, and November. Now, interesting, you go north to 2 North and then the diurnal cycle drops away. It's really strong in the first half of the year, and then it just drops away when the cold tongue comes out. What's going on? I'm going to explain that. OK. Oh. Public service announcement. We have a project, a NOAA COM project. COM, what is "COM"? Oh, I should have-- Climate Observations Monitoring project. It's titled "Flux products, long time-series, and diurnal cycle metrics from OceanSITES moored buoys as a baseline for climate monitoring and assessing models and satellite resolved air-sea interactions." That's the name of this project, and I'm leading it. And Dongxiao Zhang and Kevin O'Brien, Ken Connell are co-PIs. Nathan Anderson is helping out. Nathan is a co-chair of the DMT of OceanSITES DMT data management team. Anyway, our goal is to get all of these flux products, long time-series, and diurnal cycle metrics and make them publicly available through an OceanSITES products page. And so you, as a modeling center, can go and press some buttons and get these diurnal cycle metrics that Dongxiao just created and use them as reference against your modeled products. So we're working on it. We've computed these diurnal cycle metrics. We have the long time-series and the flux products from at least the NOAA OceanSITES available through an OceanSITES GDAC. And we have a test website that we're developing, and so we're working through this. And it will be good when it's done. It'll be really helpful, I think. It will really help the OceanSITES data sets, make it much easier to use. OK. Back to the seminar. How do we understand and predict these patterns of diurnal cycles? Let's do some math. OK. So we can use the cool skin effect. We can estimate it using the COARE flux algorithm. But it's really complicated, and I couldn't figure out how to do a standalone cool skin calculation. And so looking through the literature, I found this Zhang et al. 2020 paper, "Nighttime cool skin effect observed from Infrared SST Autonomous Radiometer (ISAR) and Depth Temperatures." And they have a new cool skin temperature effect formula that's really nice. It just depends upon the winds. And, well, OK, so maybe it's not so-- well, I don't know. This is the scatter. This lavender line is this new and improved parameterization, and it's kind of surprising to me. I still don't understand it really, but this cool skin gets larger as the winds get weaker. And maybe somebody here later on can tell me why that is. Is it because the microlayer gets thinner or thicker when the winds are weaker or larger? Or is it the wind effect on the heat fluxes that is causing it? It can't be that. Because as the winds get higher, the fluxes get larger. So anyway, yeah, we get the high fluxes at the high winds. That's what's showing in the colors. Any case, I like this formula, so I'm going to use it for the rest of the seminar. OK. Now, for the warm layer effect. I've spent a little more thought on this part. And this is a little cartoon of the Fairall et al. warm layer model. It's based on the PWP model, but it's a kind of modified model. And so it just uses the heat fluxes and calculates how much heat is getting put into the ocean and then assumes a linear stratification. Any case, it calculates this stratification and gets this temperature correction and applies that to the skin temperature to extrapolate it up to the subskin temperature. And then you add the cool skin, and you get your skin layer. So what I want to know is, can we use this model to get down to the foundation temperature? And also, I want to understand, can I use this model to better understand just the patterns based upon daily average quantities? What is the net warming for a given day? So let's get into this math here. Yeah. You integrate how much heat is getting into this layer of depth, DT. That's the trapping depth. And you integrate it from 6:00 AM, which is when there's no warm layer because it's been nighttime and it's been all mixed up. So the sea surface temperature measured at some bulk layer, like 1 meter, will be the foundation temperature. And then you integrate through the day. And you assume also that this depth, DT, will be the depth where the Richardson number becomes a critical Richardson number for mixing of 0.65. And so you have to actually know how much momentum is getting injected into the layer by the winds, so you have to integrate the wind momentum equation, too. And after much algebra-- right, because you have to solve for D, the trapping depth, from the Richardson number. And you put that into the-- any case, you get this kind of a simple equation for the warm layer correction with a very messy constant in front of it. And you can get this for the tracking depth, too. It's a messy, messy constant. But then you have just the integrated frictional velocity squareds and the integrated heat fluxes. And so now, you have your delta temperature, your trapping depth, and you can calculate this warm layer correction term here. And then you could calculate a foundation temperature just based upon this geometry here. Another way to write this actually would be as the skin temperature minus the warm layer and then add in the cool skin to get down to the foundation temperature. And you can get this warm layer correction from that delta T form. Any case, here we are. Let's see how we do. So over here, we have a plot of the temperature from a saildrone as it's going from California down to the equator and then back up. This was a eight-month mission in 2017. And as we're going along, there's all these variability. Oh, is that an eddy, or what's going on? Anyway, now we can calculate this foundation temperature and say, oh, that was just a diurnal cycle. That was not an eddy. We can go over here and say, oh, this big jump here, that was an abrupt front. That is not a diurnal cycle. And you can see how this foundation kind of touches the sea surface temperature at 6:00 AM and then smoothly goes to the next 6:00 AM, kind of smoothly. And here is another abrupt front. Isn't that exciting? These abrupt fronts Jack Reeves Eyre is going to be working on these. He's a postdoc in my group. Here we are. It's warm and then we're going into the cold tongue. And you would think, if you're looking at a satellite image, that you're approaching the cold tongue. But in fact, it's like putting on your glasses and all the detail you can see suddenly, with this data where we can see five-meter resolution, that that jump of the cold tongue, the edge of the cold tongue happened in like a couple of kilometers. And so that's something that we can look at, that kind of abrupt transition and its effect on the air-sea interaction. Jack will be looking at that. OK. And we see this, though, on buoys. So here is the KEO Buoy, and it's been out there since 2004. Actually, it's not out there right now because of COVID. We weren't able to send people on a cruise in 2020. And in fact, our cruise, which was supposed to be happening next week, is delayed again because of this Omicron, Delta, whatever. Japan closed down. Anyway, so green line is the foundation, red line is the measured sea surface temperature, and the black line is the extrapolated skin temperature using the Fairall et al. COARE algorithm. And you can see that KEO sea surface temperature at 1.2 meters is in the diurnal warm layer. And there's a really large diurnal cycle there in the summers. I mean, this is 1 to 2 degrees Celsius. Yeah. We would say that that green minus black would be the N minus the foundation. Now, I'm going to come back to this later on. We're still talking here about the diurnal warm layer math. What we want to do right now is now talk about daily-- I want to integrate this warm layer model for 24 hours and look at what is the net effect of the diurnal warming on the skin temperature. Because like the chemists, they say, I don't care about the warm layer because it's just in the afternoon and the equilibration times are more like weeks, months. And so we don't really care about the diurnal warm layer. That's just subdiurnal variability. And I say, uh-uh. No, no, no, no, no. It does rectify into the daily average. It does affect the net stratification over long periods of time. And we can calculate how much it is by integrating this at least over 24 hours. So when we do that, a one-day integration, we can pull out the daily averages of the fluxes. And so then this integration becomes a mean flux times this time number. That's actually just going to be 86,400 seconds. And so in fact, this becomes kind of simplified. Right? Because you don't have to run the model. You just evaluate it with the averaged fluxes, and you can use the average wind stress, too. But when you pull out that mean values, because the frictional velocity is squared, when you pull that squared quantity outside of the integration, there's a variance term that you have to keep track of. And that variance, I'm going to call it "gustiness." But it's really the variance of the frictional velocity over that one-day timescale. OK. And so we can calculate this every day, and we can then see what the daily net warm layer is. And it's kind of a messy quantity here but not really. It can be done. A couple of tricky things, though. In addition to this gustiness, you do have to include this penetrative radiation term. And so you have to know what your trapping depth is to be able to calculate your warm layer correction, which then you need to know to calculate your trapping depth. So you have to make a first guess. And the first guess that I did was based upon the Monin-Obukhov depth scale. But in fact, it was not very good. And so I figured out kind of a-- I mean, it's really correlated. But the trapping depth here for daily average is larger, deeper than the daily average Monin-Obukhov So I did a empirical relationship and used that as the starting value and then go through, get the first estimate and then get a second estimate and use that for the delta warm. What else do I need to say here? You can only do this integration when it's stabilizing at average heat fluxes. At other times, you set this delta T to be 0. We can do it analytically. Oh. And I should say that this gustiness that I'm using is going to be-- oh, I'm going to come back to what that gustiness is. OK? I can say. I use the Pravind Kumar gustiness that he calculated for the TAO array that's just based upon sea surface temperature to give you kind of a delta wind speed and, from that, kind of a wind stress value, gusty value. But I made sure that it never went below a certain value outcome to that. OK. So how did we do? OK. So the top is the sea surface temperature hourly just for a snippet at 0, 110 West buoy. And then the red is the 24-hour running mean. And so already you can see that that running mean is going to be above the foundation temperature, which would be going along the 6:00 AM values. Right? I should have plotted the foundation temperature on this. OK. But now, in the lower plot, I'm plotting the high past sea surface temperature at 1 meter. That's in black. And then the red line is the warm layer modeled warm effect calculated with the daily average fluxes, so that's the net warming for that day. And you can see, it really kind of captures the variability here observed. And so we interpret this as the net diurnal warming associated with the daily average sea surface temperature. OK. And now, looking in under the hood a little bit more, here again is the equation that we used to get that net daily warming, warm layer effect. And you can see, part of the problem here is, when winds go to 0, then this whole thing blows up. That's another reason we have to add in that gustiness. And so the 0, 110 West buoy was a really hard to test. Because you can see in this top panel, there's times that the winds get really down to 0. And in fact, the daily averaged wind said that it was 0. It wasn't, though. It wasn't. When I calculated the winds using just the scalar wind speed and did a daily average of the time-series of the wind speed, that's shown in red and that never got down to 0. And so that's the difference between the scalar and a vector wind speed. And when that changes direction-- I mean you can imagine wind speed being eastward for 12 hours and then westward for 12 hours and the vector average would stay at 0. But in fact, it was never 0. It was whatever it was. Let's say 2 meters per second on average. And then also, there's currents there that get really strong and sometimes are even stronger than the wind speeds. So that actually also goes into the wind stress. We'll just lump that in with the gustiness. I do calculate wind stress using the winds relative to the currents. And that is really important, particularly in these places. But the currents also have variability on time scales less than a day and contribute to this gustiness. OK. That was my point. Yeah. Here is the trapping depth and here is the daily averaged of the Monin-Obukhov depth scale calculated using the daily average data. So it's really correlated, but they're deeper. OK. Here is the diurnal gustiness formula that I used. OK. Yeah. And if you don't use that gustiness, it blows up. And so the first time I did it, I got a spike for that day that it went down to 0. Yeah. 28 degrees warming. I was like, ehh, that's not right. But it became much better behaved when you have the gustiness. Just double checking, does it work at other places? Beautifully. All right. So now, what that means is that we can start playing around with daily average fluxes and start mapping these different quantities. And so here is the famous figure from Chelton et al. 2001. Here's the equatorial cold tongue in the tropical Pacific. The purple is cold. And this is because of upwelling. But you get this really sharp front that is, in fact, sometimes abrupt between the cold water and the warm water to the north. And that front ends up having tropical instability waves on it that forms these cuspy patterns that then propagate westward. And this is an ocean phenomena. These tropical instability waves, that's a hydrodynamic instability on the water on the density front there. But he saw that tropical instability wavefront in the wind speeds magnitude when he plotted the wind stress magnitude, and that is an effect of the ocean forcing the atmosphere. And so he was saying that the atmospheric boundary layer destabilizes over the warm water. It's stabilizing over the cold water and destabilizes over the warm water. That then mixes high winds from aloft down to the surface so that you get higher winds at the surface on the one side of the front. Whereas on the cold side, you're stabilizing the atmospheric boundary layer. And so you get very weak winds there. That was a hypothesis. And then there's been 20 years of debates about, was it really this destabilization or was it barometric pressure? Because you have warm water, warm air, cold air. You're going to get a pressure gradient, and maybe that is accelerating the winds in the frontal regions. And I've written papers on that and measured the barometric pressure gradient, and there is actually a barometric pressure gradient and it is very complicated. But we can now calculate these. We have these daily average heat fluxes, and we can calculate the buoyancy fluxes into the atmosphere and the buoyancy fluxes into the ocean and the Monin-Obukhov depth scales, the stability scales, in the atmosphere and the ocean. So the bottom panels show the stability scale for the atmosphere, and it's stabilizing and very shallow in the atmosphere and the ocean over that cold tongue. And so we have forced convection there over the cold tongue. Here's the sea surface temperature. This is all based upon satellite data, the JOFURO product out of Japan. We can calculate the warm layer, map out the warm layer effect, the net effect, over this day, which is September 3, 1999, and see that, in fact, this tropical instability waves are affecting the diurnal cycle patterns. And that there, in the cold tongue, it gets really large. The diurnal cycle I'm calculating there, it's 4 degrees Celsius. We can also do the cool skin correction using that Zhang et al. that's just based upon wind speed. And it gets larger there but nowhere near. It's an order of magnitude smaller than the diurnal warming that happens. And so here's the net effect. Its warming net effect is that the skin temperature is going to be warmer than the foundation temperature in that cold tongue region. Now, let's look at the ERA5 over here, and I guess I should have had it down here. But there's this skin temperature here SST, and here's the delta. Yeah. I couldn't even use the same color palette to make this work. So you get the diurnal warming in the cold tongue, kind of. Yeah. I guess you kind of get it, but it is really wimpy here. the ERA5 cold tongue diurnal cycle is like 0.2, whereas more like 2. It's a order of magnitude different. OK. So here's my cartoon of things, and this is based upon the Barosa et al. cartoon up here at least from his 2012 paper. I should have been citations, sorry. Let's talk about the forced convection. This is the cold tongue part. We have warming of the ocean surface, and that then causes the water temperature to have a large diurnal cycle. And I had this wrong. It's actually a large cool skin effect, but it's not as large as the diurnal warming. And so we have this pattern here of foundation temperature in temperature and the air temperature that, at nighttime, is larger than the foundation temperature. But during the daytime, it's less than the skin temperature. And that then gives rise to a large diurnal cycle in the wind speeds. But on the warm side of the front, we have heat going out of the ocean. So the ocean is warming the atmosphere, and that then destabilizes the atmosphere. And you get your mixing of air from aloft that causes the high winds. Higher winds, more mixing, and we have a reduced diurnal cycle of sea surface temperature. And wind speed has a small diurnal cycle. OK. Now, let's pan out away from the tropics, and we see that these four layers are not just a tropical phenomena. This is the summer. We're still looking at 3rd of September 1999. We see some really large diurnal cycles here where the winds are changing from being easterlies to being westerlies. And so right at that transition, the winds are weak and we get a large diurnal cycle. We also have it there in the warm pool region. And this must be in the suppressed convection phase, so it's sunny out here. And then again here, this is the high pressure. There's a big high pressure system over California cold water. And likewise, down here off of Chile. OK. And a lot of these are stabilizing the atmosphere. We're getting stabilized atmosphere fluxes in these higher latitudes as well. And so I'm wondering-- When we were just talking about the equatorial cold tongue, we were kind of-- I don't know. What is it that's causing that cold tongue? The cold tongue is caused by the upwelling of the cold water. And could it be that we really need to be looking at the Hadley Cell, both in the ocean and the atmosphere? So perhaps maybe the stabilizing regions of stabilized boundary layer fluxes are associated with the subsidence branch of the Hadley Cell that happens both on the equator and then off equator at the higher latitudes. Anyway, I'm still kind of trying to get that sorted out in my mind. But I think that it's a larger picture, and we have to look at the three-dimensional circulation really, in both the ocean and the atmosphere, to understand these relationships between the fronts and the stabilizing and fluxes and ultimately the diurnal cycles. OK. And this is the cool skin effect and the net effect. And again, cool skin effect-- oh, I just want to point out here that the cool skin effect has a similar pattern to the diurnal warm layer, but it's much smaller so that the net effect-- when you have a warm layer, it dominates. And now, if we compare this to the ERA5 pattern-- again, I had to use a different color palette, because the ERA5 delta's temperature goes from minus 1 to plus 1. Whereas the way I calculated it with satellite data goes from minus 1 or minus 0.6 to like 4, more than 4. And we can look at the different places, and it's twice as much in the observed. OK. Before wrapping up this part, I did want to go back to look at the big "but." Pretty good. The KEO and the Papa, pretty good delta. But that was because we were calculating the skin temperature minus sea surface temperature. Now, we know how to calculate the foundation temperature from the buoy. And when we do that, when we compare the buoy skin minus foundation to the ERA5 skin minus foundation, it's good in the wintertime when it's just the cool skin. But when it's summertime, it's really bad. [LAUGHING] And that's because the ERA5, it appears the warm layer correction here is not strong enough. So here's my summary. OceanSITES diurnal cycle metrics, long time-series, and flux products are coming soon. There's the website. We can use the Fairall warm layer model to extrapolate bulk sea surface temperature to the foundation temperature, although there's lots of areas involved in the-- you have to worry about the gustiness and the heat fluxes. Errors in the heat fluxes will give rise to an error in that correction factor. What else? We can evaluate this, but we need to have that diurnal warm. We can integrate it for 24 hours with daily averaged air-sea fluxes to estimate the net warming or a daily average SST, but we have to add a wind diurnal gustiness to the frictional velocity. And to the extent that the sea surface temperature front is associated with stability, on the stable side of the front, the diurnal cycle of sea surface temperature wind speed will be larger than on the unstable side. But there's a chicken-and-the-egg problem. What created that stability front? Could it be the sea surface temperature front and stability front are related to the subsistence branch of the Hadley Cell? I want to learn more about that, think more about that. And then the ERA5, the cool skin compares well, but the warm layer correction appears to be too weak. So now, public service announcement. OASIS, in the last few minutes here. Sorry. OK. So Observing Air-Sea Interaction Strategy, OASIS-- oops. I used this slide from a DBCP talk. But OASIS is going to be taking a systems-as-a-whole approach for making surface and boundary layer observations relevant to the Earth's energy, water, and carbon cycles, including their physical, biological, and geological components. And it came out of OceanObs'19 where we had all these different communities coming together and writing these papers about their grand plans for 2030. And when we looked around, it was actually the carbon community, weather community, flux community, different people for different variables, the modeling-- a lot of what we were saying were the same things. And so we need to work together. And so we formed the SCOR Working Group. Phil Browne is on the SCOR Working Group. I'm a co-chair along with Seb Swart and Christa Marandino. And we are now a UN Ocean Decade program. And so we are going to be bringing in the community, that's you guys, through these theme teams. And we're going to have five theme teams observing network design and model development. And Phil is leading that group. That group is going to define the physical biogeochemical ecological air-sea interaction processes that need to be included in the OASIS to promote a predictable ocean, a clean ocean, a healthy ocean, and a productive ocean. And we need to improve the models. What are the observations that are needed to improve the models to make these things predictable? We'll have a capacity-building and partnership theme team that will make OASIS truly global, and we have an Ocean Shots and UN Decade theme team that will think big of, what do we need for 2030? We need better satellites. We need a global network of observing these-- we need to be able to measure air-sea interactions in the Southern Ocean, in the tropics, in all these different places, western boundary currents globally. We need best practices. And then we need to have fair data, model, and OASIS products. And so we have all these different theme teams, and they're going to be bringing in community. These are open groups. So looking forward to 2030 and back to 1972. This famous photo was shot by the astronauts of Apollo 17 en route to the moon in 1972, and this photo changed our perspective of our place in the universe. And now, 50 years later, we need to change. Global warming is undeniable. We need to change our role in the universe. And so I hope that, through OASIS, that we will be able to work together to track the carbon emissions as they're being absorbed into the ocean and that we'll be able to observe, understand, and predict the air-sea interactions that influence weather, climate, and ultimately the ocean environment and ecosystem and biosphere. I hope, through OASIS, that the world will gain a better understanding of the delicate balances that governs the weather, climate, and the ocean environment and a better understanding of our role in the universe. OK. Thank you very much. MODERATOR: OK. Meghan, thank you very much to you. MEGHAN F. CRONIN: Yeah. MODERATOR: That was quite an interesting project, this OASIS. Congratulations on having got it endorsed, and our hopes are with you being successful on that. We really need observations for this air-sea interaction. MEGHAN F. CRONIN: Yes. We need observations, but we need understanding and we need to have better models. We need to improve the coupling between the ocean and the atmospheres. MODERATOR: Yeah. Yeah. I think Phil couldn't agree more with you. [CHUCKLING] That's his task. So I wonder if there is any question, any pending question. Please, raise your hands. It takes a while to warm up. While people are warming up, a question. I have some-- not so sure if I understood your approach here computing the diurnal cycle. Because on the first part, as you said, I mean, you have Papa and KEO. They are either 5 SST minus the surface temperature, I think it was. No. Skin temperature or air temperature. It was good. But then you do a different estimation-- MEGHAN F. CRONIN: I did. MODERATOR: --comparing foundation temperature, something that probably ERA5 doesn't happen. MEGHAN F. CRONIN: OK. MODERATOR: I don't know what ERA5 has, to be honest, as one of the questions. MEGHAN F. CRONIN: OK. Let's talk about ERA5 in this one here. MODERATOR: Yeah. MEGHAN F. CRONIN: OK. So this comparison was the solid line-- OK. The dashed line is the ERA5 skin temperature minus sea surface temperature. OK. And the dark blue is also the buoy skin temperature minus bulk SST, and those look pretty good. So at first, we were-- this is really good. But the thing is that the bulk sea surface temperature is at 1 meter's depth, which is well within the diurnal warm layer. And so there's just a small correction that needs to bring that bulk temperature up to the skin temperature. It's not the big temperature change that you need from the foundation temperature up to the skin temperature. Right? Where was that? Here. So we're measuring both temperature, like up here. Right? Whereas the foundation temperature, this temperature difference is quite large. This temperature difference is much larger than this temperature difference. So it looks like the ERA5 is getting this temperature difference. Now, this is just based upon KEO and Papa and the tropics. We haven't looked in the Southern Ocean yet, where we want to do that. But at least at these locations, it does look like the warm layer correction that is being applied to the ERA product, that is implied by that ERA product is like a bulk SST to the skin, not a foundation. I mean, I don't know. Maybe it's that the foundation temperature-- what you're calling "foundation temperature," maybe it's not really a foundation temperature. Maybe it's a bulk SST. MODERATOR: Yeah. I mean, part of it-- MEGHAN F. CRONIN: Have to find out what's going on. MODERATOR: Part of it is that the source of SST changes in ERA5. So at some point, you have had a SST that-- I don't know whether it can be classified as bulk SST or not. And then-- MEGHAN F. CRONIN: Right. MODERATOR: --later on, we may have this foundation, so there is not this ambiguity-- MEGHAN F. CRONIN: Yeah. MODERATOR: --in the SST product that we use. And therefore, by using the same diurnal warm layer, we may change the diurnal cycle. Yes. But Anton may know better, and I see that Anton is here. ANTON: Yes. Yes. I am here. MODERATOR: Yeah. ANTON: Well, I found this very interesting. Yeah. The bottom line of your talk is, I guess, that you need to evaluate a full profile of seawater temperatures. So it's not just the base temperature that you need to observe, but you need to observe deeper layers. And if it comes to the interpretation of ERA5, I mean, there are just two temperatures. And that's what we call the "bulk temperature" and there is the "skin temperature." And the skin temperature is a modulation of the bulk temperature, if you like. It adds the warm layer and it adds the cool skin. MEGHAN F. CRONIN: Yeah. ANTON: That's basically bigger. And now, it depends on what the origin of the bulk temperature is. And I think these products, as generated by the Met Office, they filter out diurnal cycles. So it is supposed to be a bulk temperature really below the warm layer. MEGHAN F. CRONIN: Right. ANTON: I think that's my interpretation. MEGHAN F. CRONIN: That's what it seemed like reading. ANTON: In the end, I mean, what you want to do, I think, is to have a model that generate realistic profiles in the ocean and that you also use in data simulation. Because you cannot see the data simulation as done by the Met Office now independently of the modeling. That has to be consistent. Otherwise, you get the interpretation problems as we are getting now. So I think there's lots of work to be done there. And also, there's one other comment, which I forgot now, I think. Yeah. That needs to be consistent, I think. That's the main message. MEGHAN F. CRONIN: Right. This KEO buoy intercomparison I think is a good way to look at this. And I mean, I haven't done this with the ERA product. But if you overlay the skin, the SST, and the foundation temperature, you can see if you have a diurnal cycle in your bottom foundation temperature or if that foundation temperature kind of smoothly goes from the 6:00 AM pool part to the next 6:00 AM. That would be an indication that it truly is a foundation temperature. So just simply doing this type of a plot could help answer those questions of whether the reason the delta temperature from ERA5 is so much smaller than what I was showing. Is it because the ERA5 warm layer model or is it because the foundation temperature is actually a bulk SST? We could look at that and see, if there's a diurnal cycle in that ERA SST product, then it's probably a bulk SST, not a foundation SST. AUDIENCE: Unfortunately, we can't do that because we only get one field per day from the Met Office. MEGHAN F. CRONIN: Ah. AUDIENCE: So there's no variability in what we are given. MEGHAN F. CRONIN: That's hard. AUDIENCE: Although they do have products now, but it's just a model on top of that. So it's not really observationally derived. AUDIENCE: There is also the additional complication that ERA5 uses two sources of SST. One is the HadISST. That plural is defined as bulk temperature even if we only have a daily mean or even a five-day mean. And then we have closer to real time. I mean, we have the OSTIA SST, which is a foundation SST. So it's going to be difficult to do this comparison, because we treat two SST products as if they were, in a way, bulk SST while they may be different. AUDIENCE: Hi, Meghan. Tao here. For instance, we just want to add what method I would say about the switch of producting ERA5. So I'm looking at a ERA5 paper. It says clearly, after September 2007, a switch to OSTIA SST as a foundation temperature. So the parameterization skin is the same. The cool surface and the warm layer, using the same parameterization but the product had changed since 2007. So in your comparison, I think it really depends on which period you will find against. You may see the result would be different. Because for OSTIA, after 2007, it's definitely a foundation temperature. And we only have one daily mean temperature fitted into ERA5, as I believe. MEGHAN F. CRONIN: Well, we're looking at 2014 through 2020 at KEO and Papa here. This is Bo Yang who did this. So that would be in the time period that it should be a foundation temperature, not a bulk SST. But it's looking like it's a bulk SST. The ERA5 is too weak. AUDIENCE: Then the question could be-- I don't know whether it's for us-- ERA5 uses a warm layer parameterization on the cool skin. But the warm layer parameterization, I don't know whether it's based on assumed foundation SST as input or bulk SST as input. That's one question for us. The other question for you is, how confident are you in your model of the warm correction? I mean, why should your model be better than ours? When I say "ours," it's Anton's. But why should it be better? If both of them are models, where is the anchoring to observations? MEGHAN F. CRONIN: Yeah. Right. Well-- Right. The skin temperature that I have at KEO and Papa is an extrapolation based upon a COARE algorithm. But-- But-- It is-- I mean, if we're just using the diurnal cycle of the bulk SST and then-- OK. So I want to go back to that KEO slide here. OK. OK. So the red is measured, and so we're pretty confident about that. The black is extrapolated based upon the COARE algorithm, and the green is extrapolated based upon the COARE algorithm, too. But we know that, at nighttime, the bulk is measuring foundation. So we could actually just draw a line from like here, from 6:00 AM to 6:00 AM to 6:00 AM to 6:00 AM. And that would give us some estimate of the foundation temperature, and that's probably what you're getting. And so that green line, that matches that kind of common sense definition of a foundation temperature extrapolation from day one to day two. AUDIENCE: OK. MEGHAN F. CRONIN: Sorry. Did that make sense? AUDIENCE: Again, if I can make a comment? MODERATOR: Yeah. AUDIENCE: Yeah. Hi, Meghan, for interesting talk. This is John Benoit here. We have done some investigation in-house because we had problems with skin. Yes, the cool skin parameterization. But part of this investigation kind of look a bit at the warm layer parameterization we use currently in the ERA5. And I found it was a bit on the less reactive side than I would expect from the system. And the reason I'm saying that, within our code, we have as an option but it's not the used operationally to implement an ocean mixed layer model. In this case, it's from Peter Janssen which allows us to follow the first few meters under the surface and properly look at the temperature distribution. It's supposed to mimic a bit what will happen if we had a full 3D model underneath. And I found that, if I implement this in our analysis system, it produced warm layers that are a bit more reactive than our current parameterization. So your assessment might be indeed confirming what I've seen. MODERATOR: I guess still we have models. So the range of the diurnal cycle that you showed, especially, Meghan, over the cold tongue, there was a huge disparity between your estimation of maximum 4 degrees and ERA5 estimation. That it was more maximum 1 degree or even less. MEGHAN F. CRONIN: I think it was this one. Yeah. MODERATOR: Yeah. Yeah. The panel B and D. So there is a huge discrepancy there. Yeah. MEGHAN F. CRONIN: Yeah. MODERATOR: I wonder, do we have-- but you said that it was based on observations? MEGHAN F. CRONIN: OK. So the left panel is based upon satellite. Let's see. I used this equation. Right? So I was mapping this delta T warm and then using the JOFURO to get the net surface heat flux and the wind stress and from there get the oceanic frictional velocity, add in the gustiness, compute the penetrative radiation component from first estimates of the-- anyway, these are all based upon satellite estimates to get this delta T warm. And this is then interpreted as a daily average net warming, and that's what I'm showing here on the left. OK. AUDIENCE: So it's satellite observations, but then these are the formulation. MEGHAN F. CRONIN: Right. AUDIENCE: Which may be different from the one the IFS uses. MEGHAN F. CRONIN: Oh, definitely. Yeah. AUDIENCE: Can I comment on this? MEGHAN F. CRONIN: Uh-huh. MODERATOR: Please. AUDIENCE: These, I call them the "Dudley Chelton" type plots. MEGHAN F. CRONIN: Yeah. AUDIENCE: And that's related to horizontal temperature gradients. I mean, that's the main theme in all these papers. MEGHAN F. CRONIN: Right. AUDIENCE: The horizontal gradients have influence on the surface stress. MEGHAN F. CRONIN: Right. AUDIENCE: You explain it very well. There are two hypotheses about this, and that the first one is that the horizontal temperature gradient here behaves like a diurnal cycle over land, if you like. At night, the wind is low. And during the day, it's high. And that has to do with the stability and the mixing, and that's-- MEGHAN F. CRONIN: Right. AUDIENCE: This is a Lagrangian interpretation of that. That's one theory. The other one is that the data rates, data has changed diurnal cycles in the pressure gradients. I don't know the answer to that. I'm inclined to believe the boundary layer explanation with stability. But in that case, it's very important to have sufficient horizontal resolution to resolve these gradients. Because the atmospheric temperatures, they adjust to the SST. And how fast does it do that? And let's say 50 kilometer or 100 kilometer atmospheric model, can it simulate that? I mean, one of his papers also demonstrates very clearly that it depends on the SST gradient, the resolution of the SST products that you use in your model, that you get these wind modulation right. MEGHAN F. CRONIN: Yeah. AUDIENCE: There are lots of questions there that-- MEGHAN F. CRONIN: There are. AUDIENCE: Well, not now, at this stage I think. MEGHAN F. CRONIN: Yeah. And I mean, what the saildrone mission was showing is that this looks like it's a gradient that's over 2 degrees of latitude here. But what the saildrone was showing was that, in fact, it gets compressed to be an abrupt-- it's like a wall sometimes, definitely in the trough of the Tiwi. And so it's just a big jump. First year, in the warm water. And then suddenly, you're in the cold water. It's not this kind of smeared out-- and so that's a very different kind of projection on the atmosphere than this type of field here. So, yeah, lots to learn. AUDIENCE: That's an interesting test to see whether the higher resolution models are able and higher SST products are able to-- does it manifest on a higher diurnal cycle? What matters is resolution. Say, in the IFS, we will see different amplitude of the diurnal cycle at increased resolution. Correct? MEGHAN F. CRONIN: Mm-hmm. AUDIENCE: That could be interesting. The alternative-- yeah? AUDIENCE: I think that's correct in principle. In practice, the IFS handle estimates quite heavily. The diurnal cycle of the wind speed over land, and there's a reason for that, which has-- well, that's a complicated story. It has to do with geography, probably. But that's also the reason that it would underestimate this particular effect, but there is also a resolution aspect to it. And in particular, it's the stable band layer that's not very realistic for wind. So that needs improvement as well. AUDIENCE: OK. Anton, then a question for you. I mean, Meghan is also using a model. Do we have reasons to believe that the model that Meghan is using is better than the one that you formulated? ANTON: Well, I cannot answer that. You shouldn't ask me, of course. But no, no. My view is that these models do all similar things, probably. And they are all to some extent optimized with data. And so you end up with a situation where you compare models, but you basically compare the reference data that you used to optimize that model. So I think, again, emphasizing the importance of the observations. Because I don't think we have all the mechanisms in place in a quantitative way that we can build a full model from basics, and especially in the ocean actually. Because in the ocean, the mixing process is so complex. I mean, there are other currents, which I'm not sure are there in ERA5. There are the waves, which generates-- well, I mean, I'm not an expert in that at all. But that generates all kinds of effects on the turbulent mixing and the Langmuir circulation. And it is very, very complex. And I'm not sure we have all the ingredients in a really quantitative way to build a model and then put all the components together and then have a realistic model that's calibrated and it's probably not. So in the end, you are, you are left with a situation where you need your observations to get reasonable quality. MEGHAN F. CRONIN: Yep. AUDIENCE: That's where OASIS comes to the rescue. MEGHAN F. CRONIN: Yeah. Right. And in fact, I mean, part of OASIS is finding those gaps and then advocating for process studies that would help us.