 So welcome back everybody to this afternoon's session. So sorry for the small delay due to small technical reasons. So we are going to proceed. I don't know if the second speaker or the first one if Francis Dejeune is around, because we could not identify him. If it's not the case then we will ask the second speaker, Nana Cruz to share her screen and then proceed with her presentation. And she's not in meeting. Nana is not around? No. Okay. Francis also is not there. No. Okay. Any of the following ones? Semi. Semi. Yes, I'm around. Okay, good. Then you can share your screen and then proceed with your presentation. Okay. So you have 20 minutes talk and then try to manage five minutes, the five remaining minutes to all discussion. So in total 25. Please, all the guests will be good to mute your microphones. Okay. Can I start? Yes. My name is Adiso Semi. I'm assistant professor in computational data science program at Adisa University. Sorry, Adiso, can we see your face please? Okay, sure. It's much easier to follow a talk with the face to the name. Okay. I'm just, okay. Can you hear me? Perfect. Thank you. Okay. My research interest is to study the organization of convection in the organization of the mechanism of organization of convection in tropical region. So today I'm going to present to you my recent work on relationship between precipitation extreme and convective organization inferred from satellite observations. Since I will use convection several times, let me define what convection is for people who are not from atmospheric physics. So convection refers to any motion driven by the buoyancy and in the atmosphere is manifested as a vertical motion of air parcel due to atmospheric instability driven by surface heating and relative cooling of the atmosphere. Convection refers as deep when parcel originating is a watery layer exceeds the trade in virtual layer. So as you can see it in the graph in the picture, we see deep convective clouds. And this deep convective clouds are aggregated together, and therefore cloud clusters so whenever I'm talking about organization of convection, please try to picture this image, this image. So, deep deep convection exhibits a wide range of spatial and temporal organization, and they are ubiquitous in tropical in the tropics. As you can see it here in infrared image showing the inter tropical convergence on it is it, and that circles the globe around the equator. It is indicated by bands of deep clouds. Deep clouds which are quite in the image. This has been related to variability of weather and to the occurrence of extreme rain rainfall events. So observations and numerical simulation have been used to investigate role of convective aggregation in climate. So, our question is, how is organization of convection related to precipitation extremes. Up to now, there is no consensus among modeling studies enhancement of some some simulation found enhancement in extreme precipitation when applications become more organized, while others could not find an increase in precipitation extremes with organization. Here we are going to use observations of tropical and apply a new diagnostics diagnostics to link between convective organization and aggregation. So aggregation of convection and extreme precipitation. Here is the outline of my talk. First, I'll talk about characterization of precipitation field, then spatial distribution of deep convection in satellite observation. I'll discuss about the link between organization of deep convection and extreme precipitation in local and domain range. Here, latitude and longitude and latitude domains of 10 degree by 10 degree are considered to be mesoscale domains. Since this size is comparable to the domains of most cloud resolving model simulation, model studies of convective organization. In the figure, we are seeing a brightness temperature TV is brightness temperature. And the other one is precipitation field. So, both of them are this snapshot is taken at the same time. So, for a given brightness temperature we have the corresponding precipitation field. So, within the 20 degree south and 20 degree north tropical based, we consider a one degree, a one degree of a moving box of 10 degree by 10 degree. So the, this box is moving every one degree. So we have a brightness temperature for each one degree and precipitation field. So, we have like a total of 14,400 boxes for a given snapshot in the in this tropical belt. So, when we characterize precipitation and fit. First, we obtained the data from trim 3B 42 product, and this has a special spatial resolution of 0.25 degree by 0.25 degree. We consider it is this mesoscale domain, which is 10 degree by 10 degree. And the total precipitation PT is a combination of precipitating region PR and non precipitating region. So we have precipitating and non precipitating region and precipitating region further divide classified into weekly precipitating region and strongly precipitating region. According to this paper. We, we consider strongly precipitating region as whenever they have value greater than two millimeter per hour. So, we define here PT and PR to be PT is the total domain scale and PR is the local scale precipitation respectively. So we can write PT as a product of fractional area of precipitation region and intensity of precipitation region and the precipitation region can be written as a sum of weekly precipitating region and strongly precipitating region. So AR shows the fractional area covered by weekly precipitating region. AR is the total precipitating region. When we say AW is fractional area covered by weekly precipitating and AS is fractional area covered by strongly precipitating region. So, there are, there are more than 546 million mesoscale domains 10 degree by 10 degree domains when we consider all three are the data over a period of 1998 to 2000 day. And we remove, we remove all the domains for which more than 1% of the area is covered by undefined precipitation data. So, when we do this, this reduce the total number of 10 degree by 10 degree domain by 20 27% so here we are showing the PDF, the PDF of total precipitating total precipitating and precipitating region. This is the domain scale precipitating. This is the local scale precipitating. So the dark light that it is clearly shown that the local scale precipitating, precipitating precipitation is much higher than the domain mean precipitation. Then from this we calculate the 99th percentile. So in this paper the 99th percentile, the values greater than the 99th percentile are considered as extreme precipitation. So the domain as a domain scale as a domain is domain scale precipitation, which is PT has a 99% of 1.18 whereas the local scale. 90 precipitation 99% of the local scale precipitation is 4.25. So all values greater than 1.18 is considered as extreme precipitation for the domain scale for the domain scale, whereas all precipitation greater than 4.25 are considered as extreme precipitation for the local scale. So their value is PT 99 and PR 99 respectively. So whenever we see PT 99 it refers to the domain scale extreme precipitation and when we are seeing PR 99 we have to we have to connect it to the local scale extreme precipitation. So this precipitation extreme. So what we are showing here is a frequency of occurrence of the two scales. The first one is the PT, the domain scale, the lower one is the local scale extreme precipitation. So what we are seeing is for a given grid, we are calculating the frequency of occurrence of extreme precipitation in the direction of time. So most of the domain scale extreme precipitation events occur over warm pools of tropical and western Pacific Indian Ocean where one pool of tropical western Pacific and Indian oceans. Whereas the local scale extreme precipitation mostly occur over tropical. The other important parameter is number of convective centroids. So from this brightness temperature field. We identify the local minimum, like for a given three degree, three by three grid, a grid set pixels. First we identify as a local minimum and once we identify the local minimum we compare it with value with the threshold value. And in our case, 240k is considered is our threshold value. If it is less than the threshold value that point will be considered as our convective centroid. As you can see in the figure, all these great points are the convective centroids. One thing here, here we have the same number of convective centroid, but their special distribution is completely different. This one seems more organized than this one. Similarly here, the number of centroid is 105 in this, and here also is 104, but this convective entities are distributed differently for these two cases. So in order to quantify the different spatial distribution, we introduce a matrix called I work on organization index. So what this index is do once it will calculate, it will calculate the nearest neighbor distance between the convective centroids first. So every time it will calculate the nearest neighbors distance, then we'll take the cumulative density function of this distance. So we have NNCDF, which is nearest neighbor cumulative density function. So this curve shows the nearest neighbor distance and we have the corresponding nearest neighbor cumulative density function. So the black curve shows the one that we calculated. The blue curve is an assumption that it is calculated from the theoretical value of Poisson's distribution. It assumes that the distribution is random. So we use randomly random distribution. We compare these two curves. So the one which is actually calculated the NNCDF and the other one is Poisson's nearest neighbor cumulative density function. So what we found is the blue curve shows if the distribution of the convective centroids are randomly distributed, we would have found the curve to lie on the blue line. So that shows the distribution is randomly is random. But if it is below the blue line, we say that it is regularly distributed and if it is above the blue line, then it is considered as more organized. So to quantify this value, we integrate the area under the black curve. So it will give us a specific value that enable us to quantify the level of organization. So we applied this for all the domains to calculate. So all our domains have number of convective points and we can also quantify the level of organization. So here, for instance, the number of convective centroids are 34 but they are differently distributed. This one appears to be more organized than this one. So our index captured this information correctly. And we introduced this in our previous paper with Adrian Tonkin's, we applied it in the cloud resolving model and it successfully captures the evolution of the organization. And now we are applying it to the observation data. And here also we have like the number of convective points are the number of convective centers are 104 and the distribution is looks like randomly distributed. So as you can see it here, the curve seems closely overlap to each other. So, so this is our space, the organization in the level of organization and the number of convective points. So, for a given, for a given snapshot, we have for a given brightness snapshot, we have the corresponding precipitation. So we will get the number of convective points here, and we have the corresponding precipitation value from the precipitation field. In this order, the number of convective point is inversely related to the organization, but it's not always true but usually when we have more number of convective points that it is, we found to be less organized. So here also nicely captures the information. Then we study the link between the mean precipitation and convective organization. So we for each group of N, so we have a group of we classified our data into a group of number of convective points. So for each group of N, we have level of organization and corresponding value of mean precipitation. So this the level of organizations are indicated by by the dots like the blue one shows weak weekly organized the green is is less organized and strongly organized and highly organized is dark, dark rate. So, for a given number of convective point we have this different category, it is classified into different quartiles, and we can see also their corresponding precipitation value. So, for both total precipitation, I mean for domain scale precipitation PT and for local scale precipitation PR. And as we can see here in both cases, as the number of convective, as the number of convective centroid increase, the precipitation also increase. So, we see that stronger clustering is associated with weaker mean precipitation. So, yeah. Now, let's see the link between precipitation extremes and convective organization. So, for a given, for a given. So, this is similar to the previous one, but what we are showing here is the extreme precipitation. This is that. Sorry, allow me to interrupt you just five minutes left please. Okay, okay. So, this is the domain scale x extreme precipitation, and this is the local scale extreme precipitation. So what we are seeing here is for a given in, there is no systematic relationship between extreme precipitation domain scale extreme precipitation and level of organization. So it's not clear in this case, but when it comes to local scale extreme precipitation. It increases systematically. So that with the increase of organization, we see higher efficiency of precipitation. So it increase systematically with a degree of convective clustering. We further see this with the fractional area of precipitation. So, here, we see the fractional area for the domain scale precipitation, and this is a fractional area of local scale precipitation. So, the AR also increase with the number of convective point, but when we see the strongly precipitating region, a which is defined by AS fractional area of strongly precipitating region seems to be related with level of organization. Higher level of organization seems to have to be linked with higher values of AS. So, this is true for both cases for the domain scale and and for the local scale precipitation. So, the fractional area covered by heavy precipitation increases with level of organization in all the missile scale domains where extreme events occur. And then precipitating area has a stronger weights on domain scale precipitation PT than the small portion of the domain covered by a intensive brain rate. So that's why we were not able to see the link between domain scale extreme precipitation and level of organization because then then precipitating area has a stronger weight. So, in conclusion, extremes in domain scale precipitation, which occur mostly over the ocean warm pools primarily depends on the total number of convective centroids within the domain extremes in local precipitation, which occurs mostly over land depends on the degree of convective clustering. Observations are just a strong link between the intensity of extreme rainfall at the local scale and organization of deep convection, especially over land. So this is our key message. So we are, we can see it in the observation that there is a link between level of organization and extreme precipitation and at the level of local scale, which is important information especially for the modeling. This enables them to predict better if we understand the mechanism. So, finally, I have, I have done most of this work in collaboration with San Dream Pony when I was doing my postdoc at CNRS and LMD lab and also I would like to thank Adrian Tomkins from ICTP who helped me to develop a reliable organization index and this is my current institute. Thank you. Good. Thank you for your very nice talk. The talk is now open for questions. Any question? Yeah, Ali. Thank you for your very interesting talk and for injecting some climate physics into this very diverse meeting. I have two questions for you. The first one is in regards to, you know, these convection pictures that you showed. I assume that these are, they're dynamical processes, right? Yes. So these things change over time. So, in some of these density plots that you had, is this something that you get by averaging over a certain amount of time or is it a, yeah, how do you build that figure basically? So, what we did is for a, for every snapshot, like, so we calculated for a period of 1998 up to 2010. So, we consider all of them in the same books. So we put the data in the same books and we do the calculation from those data. But what is, okay, so you, you, you lump a lot of data over many, over a long times into the, into it. What is, how much, what variation is there in those patterns? I mean, there, there, obviously, there are different variation like journal variation, seasonal variation on one. So, we will consider all these things. Our interest here is just to see how they are related, like for a given, a given period, like we have the brightness temperature, and we have station one. So, how are they related? Is there a connection between the level of precipitation and the organization or the characters? This is similar to what we were doing in the modeling. So, we wanted to see how they are, how these two things are related. I see. Okay. Okay. Thanks. Thank you. Yes. Any other questions? You can write in the chat or If nobody asks, I will. Before. You had two questions. So I was, yeah, I was waiting for someone else to. So, is there, okay, I'm, I'm a completely ignorant to this, but is there any correlation between these extreme precipitation events and pollution, so aerosols? Like what is there any connection or is it completely meaningless to even ask that question? Actually, it's a different area. So I didn't look at the impact of the pollution. I mean, there are people who are working on the, on this aerosol impact. I can't give you accurate answer for this. Okay. Thanks. Good. I would have one question. Thanks for the nice talk. So you see a clear difference in between tropical convection over land to over oceans, right? Yes. Yes. And organization matters for the extreme precipitation over land, more than over ocean. Was the surprising to you? Yes. Okay. Yes, it's, it's surprising. Actually, our assumption was always, we thought that we always find organization enhance the extreme precipitation. So I would accept so as well. Yeah. But, but we were not able to see this over the warm oceans. So, in the warm oceans, what dictates extreme precipitation is the number of convective points, not how they, how they organized. When there is more number of convective convective entities, then it tends to have a higher level of extreme precipitation. But over land, there is a local dynamics, which will make it be more intensive. Yeah, I would also expect that the dynamics of convection over land is kind of more complex, even you can get supercells and maybe score lines more often. Yeah. Yes, you are. You're right. So, somehow, still, it doesn't solve this. So, some models found this enhancement related with organization and the others were not able to see it. So, our result is somewhere in the middle. So we are saying both of you are correct. So, what we are seeing over land is similar to those people who found higher extreme precipitation with organization. And for those who are not able to see the impact of organization on extreme precipitation is similar to what we have seen it over the ocean. Thank you. Okay. Good. Professor Tomkins, did you have a question? No, it's okay. I let it drop. It's fine. I think it's being covered. Thank you. Okay, sorry, I didn't see you. It's just now that. No, no, it's okay. I don't want to hold things up. It's okay. Thank you. But we have some time if you want one, two minutes is okay for us. I was just really wondering a little bit, actually, if you thought more about the implications for climate modeling with larger scale models. I mean, is this something that you think should be incorporated somehow into global models of climate where really we don't really have any representation on the sub grid scale of convective organization but they do manage to reproduce organization, should we say on the result very large scales. So do you think, because when you talk about the organizational index, you haven't mentioned very much, whether it's the smaller or the larger scales that dominate. So I was just wondering if you could talk a little bit about those scales and whether it's something you think should be focused upon in GCMs or do they already represent it. Yeah, I think the main message here is we have to understand the mechanism properly so that so that we will integrate this information in the GCMs. So we are asking to understand this mechanism in depth as it has been done in the modeling world. So we are asking for more understanding of that process to enable us to integrate this information in the GCM so that they can forecast extreme precipitation better. So, yeah, do you get your question or. Oh, sorry, I'm muted. Yeah, no, thanks. Okay. So, I think that I really appreciated your talk you can see that it drove many interrogations so now thank you very much for your talk and then we are moving now to the next speaker from Stellenburg University in South Africa. Professor Daniel McKinney.