 connecting to the server and we are now live. Okay, so welcome everybody. It's a little bit sad today in some ways because we're actually coming to the official end of this new series of seminars that Trento and I have put together. Simone is here as well actually. I don't know if you want to say something on behalf of Trento as well on the kind of closing talk, just a word of hello and from the Trento side. Yes, so from the Trento side it's not quite the last seminar. We have I think a couple of more seminars so stay tuned for the announcements but it was great. It was really a great seminar series so thank you so much for everybody for participating and Adrian of course for co-hosting and getting this started. Okay, so with no further ado this week's speaker we're certainly going out with a huge firework display because we're really very pleased to announce we have Professor Sandrine Bonny from LMD. She's a Director of Research in LMD in Paris and I was actually looking before the talk at the awards medal. She's been heavily involved in the IPCC process in the past. She has an amazing array of experiences in terms of we've seen in various conferences pictures of her flying micro lights taking measurements in clouds in the boundary layer. She organizes international observation campaigns. She is involved in huge modeling studies and model development. I remember back in my PhD days reading with Orr some of the parameterization developments she was writing an authoring for the LMD model in terms of cloud parameterization. It certainly influenced a lot of my own work in the area and so of course she's got a rich array of awards such as the Bernard Horwitz Memorial Award from the AMS. I see she was also recently been awarded the silver medal of the CNRS. So today we're going to hear actually about some of this observational work that's been going on recently and again not only is it amazing to organize a massive observation campaign but to do it in the middle of a pandemic. I'm amazed how you managed to bring this to a conclusion. So before I eat up all your time I'm going to pass over to Sandrine and say thank you so much for joining us to give this last seminar in the series. Thank you. Well thank you very much Adrian for this nice introduction and good afternoon everyone. It's really a pleasure to be with all of you. I'm sure it would be more fun to be all together in the same room but the good side of the Zoom meeting is that we can be different places together at the same time so there are some advantages. So I'm going to try to share my screen. Let's see if it works. It's not the right one. Okay. Okay. I'll try something else. Sorry. Does it work? Can you see my screen? It's perfect. Yeah, great. Okay so yeah it's... I'm very glad to be able to share with you some of my excitement for the topic that I find really interesting and I hope that you will be convinced that it's very interesting at the end of this talk. It's about the mesoscale organization of clouds in the trade ring regions. You'll probably know that the organization of convection has raised a lot of interest over the years in the community and Adrian is one of the responsible for that. One of the people who raised a lot of interest for this some time ago. And many people are interested in the organization of deep convection but their shallow convection also can organize in different ways and it's at least as interesting as the organization of deep convection I would say. And so I'd like to talk about the mesoscale organization of clouds in a particular type of regime which is a regime of the trade winds. And this type of organization has not been studied very much so far but it's in the trade ring regions that we organized the Uri Kaffir campaign last year as Adrian said. And it was not during the pandemic but right before the pandemic because the campaign finished in February 2020 so just before the start of the pandemic. And so as part of the preparation of this campaign we looked a lot at satellite imagery because we wanted to know what kind of situations we could find during the campaign and that's how we got really interested in this issue of the mesoscale organization of clouds in these regions. So I will tell you what we've learned during the preparation of the campaign and the subject and also what we've started to learn from the campaign itself and obviously it's just preliminary findings because the data processing of the campaign is still ongoing so it's very preliminary stuff but I hope it will at least raise some interest for the campaign. So I'd like to mention first that the work I'm going to present obviously has been done in collaboration with many people mostly from my group but also from people in Germany and particularly with Björn with whom we have organized the campaign together and many young scientists. So when we think of the organization of shadow convection most of the time we think of particular types of organizations and one type which which comes to mind is cloud streets or roads which have been studied for a long time that we can find over the high-latitude oceans but we can also see this type of organization overland sometimes. Another type of organization which has been studied a lot in the past is this type of mesoscale cellular convective systems so I'm going to remove this. So this type of systems can be really spectacular if you look at them in satellite imagery and you can in this picture recognize what we call usually closed cells or open cells depending on whether the clouds are surrounded by clear sky or whether it's clear sky which is surrounded by clouds which depends on the proportion of clear sky and cloudy areas and so the open cells and closed cells are often found close to each other sometimes you have open cells in the middle of closed cells as you can see here. Well so it's a really nice type of organization and there have been many studies already on this topic in the past but and and in particular so in this particular study by Mulbauer they try to recognize these patterns automatically by using a neural network so that they could screen at many satellite images and recognize this type of patterns of closed cells or open cells and then they could make a climatology of these patterns and what they show is that the closed cells can be found mostly over the mid-latitude oceans but also at the eastern side of the ocean basin here so where the ocean is not warm. The open cells are less frequent than the closed cells but they can be found a little bit more widespread as you can see but what I find really interesting in this study that actually the show that over most of the tropical ocean and in particular in the subtropics where it's red we can't find this type of patterns so this type of organizations are very spectacular but they are not representative at all of what's going on in most over most of the tropical oceans and in this study there the type of organization that they find in those regions the name in disorganized MCCs as for disorganized organization if you like because it does not correspond to the classical pattern of closed cells or open cells that has been found as well so the the only cafe campaign that we wanted to organize was going to take place in this area here of the western tropical Atlantic close to Barbados so right in the middle of a red spot so we we wondered what kind what kind of organization we could find there and so in as part of the preparation as a campaign we decided to look at it in more details and so what we did is that with a few colleagues we sat in a in a room for several days looking at satellite images on the big screen for several days and and trying to to detect some prominent patterns of organization and we found that the organization in this area so this is Barbados here this is the Caribbean islands in this area the middle scale organization can be very valuable from one day to the next for instance here you have a day during which clouds organize along lines or arcs like here which are reminiscent of corpus but on on a few days later on on on another day you can find this kind of organization where here it's completely different you can it's still shallow convection and the cloud tops are around 2.5 or 3 kilometers but as you can see here for instance we have clouds that are much more extensive and in particular there is some stratiform anvil at the top of the cloud which can be a several hundred kilometers large separated by very clear sky situations so by looking at many images several hundreds we pointed out the fact that there were four prominent patterns of of cloudiness that we could find in this region and we name we gave names to this pattern because it's much easier to talk about them once you have a name to talk about them so the first type of organization we named it sugar because it really looks like sugar powdered over the ocean and it's it's associated with very shallow clouds very small clouds that you can hardly see from from satellites another type of organization we named it gravel because of the structure the texture of the of the image and it's associated with shallow clouds but also some deeper clouds and another type of organization we named it fish because it looks like the skeleton of a fish it's composed of large clusters of cloudiness with some smaller scale structures within it and the last type of organization we named it flowers because it why it corresponds to the the type of organization we were looking at earlier and that is it's really a beautiful one with associated with the cloud systems with a stratiform and zero so we can look at this so this is now should I say that but the each image here is a domain of about 1000 kilometers by 1000 kilometers so we can look at this patterns from space and we can also look at the clouds that compose these patterns by looking at ground based observations in particular we used the radar that that was installed in the Barbados cloud observatory to look at the cloud types that were found within those patterns and by doing this we we found that there's different patterns associated with different cloud types this sugar type of organization as I said earlier is really associated with those very very shallow and thin clouds that hardly exceed the level of conversation this gravel type of organization is associated also with the presence of this very shallow clouds but also with deeper clouds that can go to three or four kilometers and that can rain really sometimes and and those two types of organization are associated with more extensive cloudiness especially near the inversion level and this one as you can see here very clearly is associated with an angel very thin layer of stratiform cloudiness so this patterns first were identified visually so it was really subjective and but afterwards of course we wondered whether it could be possible to recognize as different patterns in a more automatic way and there are different ways to do that so there were some initiatives to to do it through machine learning and if you want to to do it through machine learning you need a training dataset so what Stephen Rasp told is that they developed a cloud sourcing platform so that many people could contribute and look at satellite images classify the patterns and and then we have a huge classification dataset that can be used as a training dataset for deep learning algorithms and there are different techniques of machine learning that can be used now to to recognize these patterns and in parallel we wondered whether it was possible to recognize as different patterns without machine learning but just using a simple methodology to analyze the satellite images and that's what I'm going to tell you a few words about it because we are going to use it for for the next using that I will show you so here we use two stationary satellite observations in the infrared for 20 years and we try to characterize a special organization of the clouds by using two different metrics one metric is a mean size of the cloud objects of the cloud clusters that we can detect and the other metric is the so-called iorg organization index that was proposed by Adriana which years ago with Adisu semi and that characterizes the distance in between clouds so it's based on the PDF of the distance between the nearest neighbor's clouds so we we did that and we so each we took many days in winter and for each day we we associated one point in this 2d space with a associated with us two metrics so we have a big cloud of points and we try to differentiate four extreme cases of organization so we try to to separate the most contrasted types of organization by considering the first and third size of each metric so that effectively defines four quadrants that we can name abcd right and then we try to see whether there was some relationship between these different quadrants and there the four prominent patterns of organization that we had found visually and that's what we show here we found that this this quadrant for instance was largely associated with the flowers situations but we had identified visually and this quadrant here was associated mostly with sugar type of situations and the two other quadrants were a little bit more ambiguous but at first order they were associated with the predominance of fish or or gravel type of situations so then we renamed this four quadrants as flowers fish gravel and sugar and use this simple method to to recognize the patterns from satellite observations and then we wanted to see whether there were there was some relationship between the cloud patterns and the environmental conditions in which they occur so we consider different environmental properties or variables like the ssd the lower trapezoid stability the wind speed the wind shear etc etc and among those different variables we found two of them that turned out to be most discriminating for the cloud patterns which are the the surface wind speed and the lower trapezoid stability so typically we found that so this is the lower trapezoid stability or the strength of the inversion the level of the trade inversion this is the surface wind speed we found for instance that the flower type of organizations were found mostly when the situation was very windy and very stable while the gravel type of organization was also found when the surface wind speed was strong but the atmosphere was less stable for instance so we looked at this by based on 20 years of data that we could look at it or so for individual seasons for several years and what you can see here is that depending on the year we had the predominance of one type of pattern or another type for instance during this winter season we had mostly fish patterns and it was consistent with the fight that i'm doing this year this season we had a lot of stable situations and low wind speed while on this other season for instance we had the predominance of windy situations and we had mostly gravel and flowers depending on the stability and so on so the the association between the environmental conditions and the type of patterns in prisons was appeared to be quite robust as well more recently jesse caviar it all looked at diagonal variations in in the patterns and they did this by identifying the patterns through machine learning and applying the methodology to the goes 16 satellite observations and they showed that there is also a diagonal modulation of the frequency of the different patterns and when looking at the surface wind speed and the stability they also find that there is different cut patterns are stratified by the wind speed and the stability during all the day and not just one particular time of the day and they found that the diagonal variations were more associated with wind variations during the day than with the stability variations then the question is whether the different patterns have different relative impacts so we looked at this by using the methodology i was discussing earlier to recognize the patterns and we looked at the net cloud relative effects of this pattern so the cloud relative effect is the impact of clouds on the earth's radiation budget at the top of the atmosphere and we looked at this as a function of the low cloud cover associated with each of those patterns and of course when the cloud cover is is larger the clouds have a stronger cooling effect than when the cloud cover is smaller and what we find what we see in this figure here is that at first order if the different patterns have different relative effects and indeed between this pattern and this pattern there is a factor of three briefly of difference at first order we can interpret this difference mostly as a the consequence of the fact that there's different patterns associated with different low cloud covers whether for a given low cloud cover there is also a systematic difference in the relative impact between the different patterns that could be associated with differences in micro physical properties for instance i would say that from this data set it's not obvious it's it's not significant at least but at first order it's really the macroscopic differences between those different patterns that explain the fact that takes a different relative impact so then we can wonder what are the implications of this for for the cloud feedbacks when we want to understand cloud feedbacks especially the low cloud feedbacks what we want to do is to to understand what are the environmental factors how the environmental factors affect the cloud properties right and the relative properties in particular and we do this by considering different environmental conditions and potentially we can consider different cloud types when when we try to relate the environment to the cloud properties and usually we differentiate the cloud types from the vertical structure only but what we've seen here is that there is also a link between the environment and the mesoskeletonization and between the mesoskeletonization and the cloud properties so that we can wonder if to understand this link between the environment and the cloud properties the relative properties in particular whether we need to consider the role of the mesoskeletonization this is an open issue and clearly at the moment I would say that we don't really know whether or not it's important to consider this route to understand this but that's that's an interesting issue at least so we'd like the the models to tell us something about this and to tell us whether they're mesoskeletonization of cloud matters for cloud feedbacks but we don't have the answer yet for this question first of all if we consider climate models like the semi-models obviously these models not represent the mesoskeletonization of shallow convection so they cannot help us very much on this we can consider other types of models like the cloud resolving models or the large eddy simulation models but even in these models representing the different patterns of organization can remain a challenge for different reasons and to appreciate the challenge I'm showing you here this satellite picture again on a scale of a few thousand kilometers here that shows you that at this scale you can really find many different types of patterns sugar here gravel here some flowers and here some fish and the patterns are associated with different space scales so if you have a very high resolution model that can represent mesoskeletonization and some scales if you want to to to look at the sugar type of clouds or maybe even the gravel then it's okay if you have a small domain for your model but if you want to study the fish type of organization of flowers then you need a much larger domain and at the moment most of the cloud feedback studies for low clouds which have been done for using this type of explicit models I've been using very small domains either very very small domains sometimes even smaller than that where you don't have any organization or really very small scale organizations or you have a larger domain like 50 kilometers sometimes 100 kilometers where you start having some organization like cool pools for instance but with this type of domains you cannot get fish or flowers for instance you need an even larger domain but at least we can look at what has been learned about the role of the mesoskeletonization of clouds in the low cloud feedbacks with this type of domains keeping in mind that it's only it considers only the role of some particular types of organizations not the the biggest ones so for instance in this study by Raffaella Fogger she used a large dissimulation model with and without special organization of shallow convection and she did this by using different domain sizes for the simulations and what she showed is that depending on whether she was running on this size of the main or on this one she got very different cloud fraction profiles so at first order the macroscopic properties of the cloud field really were strongly different very different depending on whether or not there was some organization on the large domain or or not and then she also made experiments in which she imposed Fogger warming at the surface and she found that in all cases the cloudiness slightly decreased and the response was much smaller in the case of the large domain than in the case of the small domain so what it shows at least is that there is a potential for the mesoskeletonization of clouds to affect the response of clouds to warming but obviously yeah those results remain quite uncertain because as I said again the simulations represent just some particular types of organizations but at least it says that the organization can really have a huge influence on the microscopic properties of clouds and their response to climate change so if we want to consider more types of mesoskeletonization and we need larger domains and ideally we'd like to have global simulations with a very fine simulation so that we can represent this different organizations and at the moment where we're starting to to see the emergence of this very high global resolution models like here I'm showing you a snapshot from the icon model developed at MPI which is a global 2.5 kilometer model and at this so it's a very short simulation and it's not a climate change simulation not yet but you see that in this simulations at high resolution the models spontaneously simulate something that is reminiscent of what we can see in observations through this type of fish or flower patterns for instance so that's encouraging and say that probably we learn a lot about this pattern and their role in climate with these new types of models but at the same time a resolution of 2.5 kilometers is a little bit coarse if you want to represent really the shallow cumulus processes because the typical size of shallow clouds can be much smaller than that so it raises other problems so it's even for those models it will be very important to understand how realistic the shallow clouds are in these models how they interact with the environment and then ultimately we hope to learn something about the role of cloud feedbacks in cloud feedbacks and the reason why we are really interested in knowing how these clouds might respond to global warming is because as you might know the trade wind clouds are suspected to play an important role in climate sensitivity well they can play an important role by being very sensitive to warming or by not being very sensitive to warming because the trade wind clouds actually correspond to the most prominent type cloud type on earth so depending on how they will respond to warming they will influence the climate sensitivity a lot even if they don't respond to warming that will be a very big constraint on climate sensitivity to know it and at the moment the semi-models as you know exhibit a wide range of climate sensitivities and between the low sensitivity and the high sensitivity models what has been shown is that the difference comes to a large extent to the response of clouds in the trade wind regimes especially in the response of clouds at cloud base so that was one of the strong motivations for organizing this field campaign that the Eurica field campaign because we really want to understand what controls the cloudiness in the trade wind regions and another question now that we we'd like to to address is whether the lack of representation of methods scale organization in these models can be an issue for for the trade community cloud feedback similarly by these models so for all those reasons we are very much interested in the trade wind clouds I was interacting with our environment and that's why we organized the Eurica field campaign with Bjorn and David Farrell from Barbados so there are several overview papers presenting the campaign one that was written before the campaign presenting the motivations and the experimental strategy for the campaign another one presenting what we actually did during the campaign that is I've just been accepted for publication so you can look at these papers if you want to know more about it but so I will present you a few main things about the campaign first so the campaign to place near Barbados here and Barbados is really a great place to to study trade wind clouds because it's here so it's the train winds are blowing like that and so the the clouds that arrive in Barbados are really not been affected by any land mass over thousands of kilometers so it's really very much like being in the middle of the ocean and looking at clouds so it's a great place to observe clouds and in addition what has been shown a few years ago it is studied by Ibrahim Medeiros and with Nyan see that the clouds at Barbados are representative of the clouds of the trade wind regions both in observations and in climate models so it's a nice place to to study this type of cloud so this is where Eurica to place Barbados is here and most of the the operations to place east of Barbados within this circle here and also along the transect more or less aligned with the the trade winds but and initially the Eurica field campaign was meant to focus on cloud processes and their interaction with the palm oil layer and so on but as we were preparing the campaign actually many other groups and and scientists found that it was a great opportunity to address complementary questions and so there there were also initiatives focusing on on characterizing the RC interactions in this region and also the the the impact of ocean eddies and ocean submerses care structures in the and the impact that they could have on the RC interaction so the and the studies developed in particular in this branch here so south of Barbados where we there are big ocean eddies but what I'm going to talk about in the next few minutes we will be mostly based on what we've we've done within this area here east of Barbados so initially the campaign was meant to to use observations from two research aircraft aloe from Germany and the ATR 42 from from France and also observations from the Barbados cloud observatory but because the scope of the campaign increases increased a lot at the end we had many more observing systems and that was great that we had four research aircrafts and we had four research vessels participating we had a big radar installed on Barbados especially for the campaign to measure precipitation and we also had a lot of autonomous observing systems to characterize either the ocean or the atmosphere or the the ocean surface to study the RC interaction and many of the systems actually were very new and some of them were used for the first time and it was really something special for for this campaign there are so many types of observing systems so based on this we had a very rich ensemble of observations for the atmosphere for the operation for the ocean surface and and we can really characterize the atmosphere and the ocean and on a wide range of scales so if i come back to the what we really wanted to do in terms of of god processes initially for this during this campaign we we designed the strategy of the eureka field campaign around this at the beginning it was really based on the and the use of this two aircraft the the german aircraft aloe was flying large circles in their petroposphere at about nine kilometer height and on on board the aircraft there was a rich instrumentation for remote sensing to to observe clouds from above in different ways and also the aircraft had the ability to drop drop zones along the circle so that we could really characterize the environment in which the clouds occur and at the same time that halo was flying this large circles in the petroposphere the french aircraft the atr but also the british aircraft twin otter were flying in the low petroposphere and with the atr in particular we we focused mostly at the cloud base level but also in the sub-cloud layer to observe clouds from from below or within clouds so one of the key measurements that we we did during eureka was measuring the divergence of the large-scale vertical motion that's because we know that there's the large-scale vertical motion is very important in controlling the the the cloud field and a few years ago with beyond we had shown that we could measure the large-scale divergence by using drop zones so if if an aircraft is flying a large circle like that and if we dropped many songs songs along the way then each song measures vertical profile of the horizontal wind and by looking by by using all the drop songs along the circle we can compute the the divergence of the horizontal wind so the the divergence and and by integrating upward we can infer the the large-scale vertical motion so the size of this circle was about 200 kilometers so what we measure is the large-scale vertical motion that what but we can still characterize as being the mesoscale the mesoscale so we we did that a lot during eureka there were three circles in the first half of the day and three circles again in the second half of the day each with 12 drop songs dropped for each circle so many drop songs every day and based on this we can really look at the vertical structure of the divergence and the vertical motion as a function of height and vertical and as a function of time as you can see there is a very rich structure in the profile of mass divergence and by comparing the the measurements on one day to the next we could see that there is also a lot of variability really of the divergence of the vertical motion so if we look at the large-scale vertical motion if we average all the measurements over the over a month over the the whole campaign we see that above the boundary layers the vertical motion is is subsiding and fairly small and that what balances the mesoscale relative cooling that's an order of magnitude that we expect but if you compare this with the measurements that we can get on one particular day at one particular moment of the day you see that the order of magnitude is very different and really the the vertical motion was extremely valuable and even in the boundary layer we could have some subsidence but we could have some large-scale ascending motion as well in the boundary layer that was quite common. Another key measurement in Eureka was the estimate of the mass flux at cloud base and there are different ways through which the mass flux can be estimated but one way that we wanted to use in Eureka was to estimate the mass flux through a mass budget of the sub-cloud layer so if you write the equation for the height of the the mixed layer which is the layer here so the time derivative of the height of this mixed layer depends on the magnitude of the last vertical motion if you have subsidence it pushes the mixed layer downward it also depends on the end-trainment at the top of the mixed layer that depends on the turbulence at the top of the mixed layer and it depends on the mass flux at the cloud base that exports mass out of the sub-cloud layer so if we measure the depth of the mixed layer we measure the mass-scale vertical motion as I explained just before using drop zones and we estimate the entrainment from the surface flexes turbulent flexes and there's other vertical the jumps in the temperature and humidity at the top of the mixed layer then we can estimate the mass flux as a residual and Rafaela Fogel has shown that this methodology could work while she did that using her large by analyzing earlier simulations but then she applied the same methodology to the Eureka observations that's what she's doing it at the moment and here I'm showing you a time evolution of the of the mass flux that has measured during the whole campaign and as you can see there is a lot of variability of the mass flux both on day-to-day variability but also within a given day there is also a large dinar variability of the mass flux then we another key measurement was the cloud-based cloud fraction partly for the reason that I explained earlier the fact that we want to really understand what controls the cloud-based cloud fraction for cloud feedback studies and and to to do this we propose while we can measure the cloud fraction in different ways again but in the in the trades the cloud fraction is very small just a few percent so we need a lot of sampling to get accurate measurements and so for this for this reason we propose to measure the cloud fraction through horizontal LiDAR and radar measurements so by using a LiDAR or LiDAR looking through the windows of the aircraft of the ATR and by using the LiDAR radar synergy we could detect the presence of clouds and finally another key measurement for the campaign was a radiative cooling because we know that in these regions there are many processes which depend on radiation for instance the circulation that some mesoscale circulation can be driven by a radiance in radiative cooling in those regions so what an Aliya Albright and a few colleagues have done is that they used all the soundings of the campaign coming either from radio sounds or from from drop sounds and there were many of them as you can see and for each sounding they computed the vertical profile of radiative cooling in clear sky so that we have this great data set now finding in terms of mesoscale organization during Eureka we we were lucky enough to sample a large diversity of cloud types and and cloud patterns so on this day for instance we had something like a fish type fish pattern here we had something like flowers like here we had gravel and on these days we had well we don't see the clouds on these images but we had mostly sugar clouds so we we we could sample the different types of organization that was nice so if you want to to remember something from Eureka actually it was really a campaign during which we were able to characterize the clouds and the environment of clouds on a very wide range of scales and through many many different types of measurements so we could characterize the mesoscale was a large scale through satellite observations through radar measurements we could measure characterize a very microscopic scale for instance some people were looking at the individual cloud droplets and how they aggregate from rain for instance within clouds there were many radars looking from the top looking sideways or upward or and there were high-resolution radiometers to look at clouds with very very good resolution and there were macro physical measurements within clouds as well measuring the cloud water content vertical velocity and so on and in addition to that for the environment we had measurements of the last kind of vertical motion we had measurements as a surface but surface flexes and much more so it really we have a very large dataset now that that can be used for a wide range of process studies if you're interested in clouds and their interaction with the environment so now I am I hope I don't know how I am in time but I'm going to to give you a few first insights from the campaign so one first lesson from the campaign that was very clear is that shallow convection generates a lot of cold pools and we see we saw many of them during the campaign you see this this arcs here the developer around this deep clouds here while deep they go up to a few kilometers only but you really see this cold pool so it's not only in regimes of deep convection that we can find cold pools but also in regimes of shallow convection they are very common you can see them also very clearly from aircraft pictures and the cold pools can be very well detected from the soundings so it's not only by looking at the horizontal heterogeneity of temperature and humidity that you can detect them but also by looking at the soundings because they are associated with a very shallow mixed layer and so some people like Ludovic Tuzepaefer they characterize the cold pools during your weekend and they showed that they were colder than the environment a bit moisture than the environment and windier and of course the cold pools were only one type of organization that were observed and and we as I said earlier we really observed a large range of organizations and what we we are very much interested in in my group is trying to understand what how what controls those different patterns of organization and in particular we know that in the trades the cloudiness is strongly related to what's going on in the sub-cloud layer particularly the presence of thermals or coherent circulations so we really wanted to to characterize this connection and we did this mostly by using data from the ATR and from the drop zones so on board the ATR we had turbulence measurements we could measure temperature, humidity and winds at 25 Earths which because we were flying at about 100 meter per second it gives us a special resolution of about four meters so very high resolution and based on these fluctuations we could detect the presence of of boundary layer thermals so we can count them we can as a characterize the size of the thermals so over Eureka on average we found that we had a mean density of thermals of about one thermal every kilometer and we found that 20 to 30 percent of the thermals were kept by a cloud so with a horizontal LiDAR radar measurements we could characterize a cloud field within the rectangle around which the ATR was flying and develop a cloud mask with a resolution of 25 meters and based on this we can compute the cloud base cloud fraction so it's a small number on average about four or five percent but what's interesting is that first of all the estimate of the cloud base cloud fraction is fairly consistent while it's still preliminary analysis and data but at first order it's fairly consistent amount of different measurements that we had on board the aircraft from remote sensing or from EC2 observations and what's interesting is that the cloud base cloud fraction as you can see varies a lot from one day to the next which was a bit different from what had been suggested before the campaign where people thought that it would be a more or less constant so here we see some some variations and because we expect the cloud base cloud fraction to be related to the presence of thermals we looked at whether there were there was some relationship between the variations of the cloud base cloud fraction and the density of cloudy thermals so this was derived either from LiDAR radar measurements or from EC2 measurements and that was derived from the turbulence measurements on board the aircraft and we do find a close connection between the two so if we want to interpret the variations of the cloud base cloud fraction then we have to look at what can modulate the density of cloudy thermals and we looked at different environmental conditions that could affect this and we found at least wait it's really ongoing work but we found at least two properties of the environment that seem to be important for the modulating the density of cloudy thermals in particular the strength of the mesoskeletal vertical velocity in the water atmosphere and the surface wind speed and the flights during which we had the more the many thermals were associated with strong wind speed conditions and large-scale ascent in the boundary layer but now if we're looking at the link with the mesoskeletal organization at first we can be a little bit disappointed because if here I featured three different days three different flights that were clearly associated with different cloud types and different patterns of cloudiness and yet if you look at the cloud base cloud fraction they are more or less the same cloud base so it seems that the mesoskeletal organization at first order is not imprinted in the cloud base cloud fraction however if we look instead of looking at the total cloud fraction at cloud base we look at the individual clouds and we look at the size distribution of individual clouds then it's very different here I'm showing the size distribution of the cloud base the cloud basis and in this case when we had mostly sugar clouds we have only a very small cloud cloud basis well in this cases we have many cases of very small cloud bases but there's also a smaller number of much larger cloud bases so we looked at this more systematically for every flight of the campaign and we found that really the size distribution can really be interpreted as a mixture of two populations of clouds one population which corresponds to many clouds of very small size and another population associated with smaller fewer clouds but much larger cloud bases so we found that we could fit the size distribution with a mixed exponential function so two exponentials each exponential representing one cloud population or one mode of the cloud population and each mode being characterized by two parameters one is the fraction of the population that this mode represents and the other parameter being the the length scale of this mode which corresponds to the mean size of the of the cloud population in this mode so we did this for every flight and this is the first mode for the different flights so on average the the length scale for this mode is is quite small it's about 100 meters on average so it's what we infer from the LiDAR radar measurements and actually it's interesting that there if we look at the size distribution of the cloudy thermals that we can infer from the turbulence data in situ data then we find that the mean size of the cloudy thermals is also about 100 meters so the way we interpret this first mode is that corresponds to the cloud that gap individual thermals and that's why we have so many and they're associated with small cloud sizes but there is a second mode and the length scale of this second mode is very valuable from when that is next as you can see here the length scale can vary from a few hundred meters to a few kilometers what we found is that this second mode the length scale of this second mode is very much related to the presence of of rain so when we have a large cloud basis is when we have a lot of precipitation and the way we can interpret this relationship is the fact that when we have a large cloud base then the clouds are associated with a weaker entrainment and therefore the clouds can go deeper and can generate precipitation more easily and what's interesting is that there seems also to be some connection or relationship between the size of the cloud basis and the different cloud types that compose the mesoscale patterns of cloudiness for instance on this flight here we had a sugar type of situation so very small clouds we had the single mode of single cloud population associated with very small clouds and in this case we had a larger cloud basis and then we had the superposition of small clouds but also a few deeper clouds and when we have even larger cloud basis then it's when we start having rain and and even deeper clouds and the formation of stratiform cloudiness near the inversion level so it seems that the size distribution of cloud basis might be important to understand how we we go from one type of cloud types to another and understand the the emergence of these different patterns of cloudiness so then the next question is what controls the size of the cloud base as we've seen the typical size of thermals is about 100 meters or a shoe at maximum a shoe 100 meters so the the large cloud bases of a few kilometers are much wider than this which suggests that they might be associated with an aggregation of thermals but then the next question is why do we have such an aggregation of thermals so it could be related to to the surface heterogeneities potentially or it can be related to internal organization within the sub cloud layer so we're starting to started to look at this using the turbulence data set again it's very much work in progress but you're looking at the spectrum of vertical velocity within the sub cloud layer for instance we find that there are typical characteristic length scales which most of the time are of the order of one kilometer which is consistent with the mean density of thermals that we found that was we have about one thermal every kilometer in these regions but sometimes we have those characteristic length scales which are much larger than that and so now the the question is what what how to interpret this and and this is yeah really work in progress so I will conclude now because I think I've been a bit too long but as I am I hope I have convinced you that the shallow clouds can really organize in many different ways and the organization that you can sign in the trade winds regions is really interesting because it can be very diverse this organization is associated with different types of clouds and and there's different organizations depend on the environmental conditions and and they can they have different relative impacts so that they could play a role potentially in cloud feedback so we are not sure about it yet thanks to Eureka we can start trying to understand what controls these different patterns of organization and our first analysis suggests that the organization of shallow convection is imprinted in the size distribution of clouds at cloud base that shallow rain mostly forms from clustering and that the large cloud bases are associated potentially with our coherent middle-scale organizations in the sub-cloud layer but it's yeah it's really a work in progress the very preliminary findings and we hope to test this hypothesis further in the future so thank you very much thank you very much Sandrine maybe so that I can see you thank you very much you were saying about the advantages of the the zoom meeting one of the disadvantages though is you can't hear everybody clapping at home I assure you there's ruptuous applause and carrying around the internet so I've got a couple of questions coming in and I've actually got one or two of my own the first person actually I'm going to hand over to is Daniel and then to Simone let me just ask to unmute okay oops sorry I clicked twice Daniel you should be able to speak now okay yeah so thank you Sandrine that there was a great talk very very interesting um I just was wondering if you had if you had the chance during Eureka to look at any aerosol impacts in particular maybe the Saharan dust that probably has some strong impact there so yes there were aerosol measurements and as you might know in Barbados is a great place to to measure aerosols and it's been historically a place with aerosol measurements especially to to look at these aerosols coming from Africa so there are several decades of measurements and during Eureka we did measure aerosols as well and we had a few episodes of dust indeed a lot of dust coming from Africa and apparently according I'm not a specialist of aerosols but according to some colleagues it seemed that in winter usually we don't have so much so many intrusions or arrival of dust aerosols from Africa but apparently this time we have more and more we have this more and more often even in winter and we had a few episodes during the campaign so yeah that can be an opportunity to look at how they affect different things okay thank you sorry Daniel um we also have a question of clarification from Menorah um so yeah so um as far the definition of the limits for the four groups of four types of clouds I'm not sure and understood properly how you defined the the limits between them so I remember that group where there was the the ABCD groups and and this gray crossing the middle to separate them to make sure that the groups were far away but how did you decide like what's A and what's gray what's nothing as an example so in our in our classification of the organizations using satellite observations we defined the four groups of organizations in a very super simple way we just look consider the first and last size of each metric right so we didn't know priori whether it was flowers, gravel or anything else we just wanted to distinguish contrasted types of organization that's all that just afterwards that we looked at what type of patterns had been classified visually for these different quadrants and that's when we we found out that one quadrant was mostly corresponding to flower patterns and and so on but at first we just wanted to to distinguish different types of organization not knowing whether it was flowers sugar or anything else but