 So, good afternoon everybody. It's my pleasure to be, to welcome and congratulate the 2021 ICTP Prize winners who are present there at ICTP today. Rondru Tiana, Bari Malala and Narendra Oja. I wish I also could be present but I have to join by Zoom. I would like to just say a few words about the ICTP Prize which was created in 1982. It recognizes young scientists from developing countries who work and live in those countries and who have made outstanding and original contributions to physics. Each year, the ICTP Prize is given in the honor of a scientist who has made outstanding contributions to the field in which the prize is given. The 2021 ICTP Prize is dedicated to the memory of Jacob Bueblis who discovered the connection between sea surface temperature and easterly winds anomalies in the equatorial Pacific. He helped towards an understanding of the El Nino Southern Oscillation by suggesting that an anomalously warm spot in the eastern Pacific can weaken the east-west temperature difference, disrupting trade winds and which push warm water to the west. So these results have been important in climate science and the winners of this year's ICTP Prize are their own field of scientific research is closely connected. I would like to now first read the citations for Rondruth Yama Bari Malala and then the citation for Narendra Orsha. And perhaps then I will hand it over to Fred to give a more detailed scientific introduction to the two winners this year. So the 2021 ICTP Prize is awarded to Dr. Rondruth Yama Bari Malala for her outstanding and pioneering contributions that have advanced our understanding of the role of the Indian Ocean on the climate of Madagascar and South Africa and its response to climate change through the application of modeling tools and careful data analysis. The citation for Dr. Narendra Orsha is as follows. The 2021 ICTP Prize is awarded to Dr. Narendra Orsha for his outstanding contributions to the field of atmospheric chemistry and physics by performing in situ measurements, satellite data analysis and chemistry climate modeling is over South Asia. Dr. Orsha, just that these have had an enormous impact on our understanding of the influence of anthropogenic and biomass burning emissions on the Southeast Asia air quality. So I will now give the floor back to Fred since he is there. Thanks very much, Adish. So we will proceed with the with the handing of the document. Okay, first. Okay, okay. Congratulations. Congratulations. Okay, congratulations to both of you. So before we start the scientific presentations, I will just very briefly not talk very much, very informal, you know, informal talk about a little bit of the history of the candidates. Also, Rondro Barimala started her career in climate science, at least when she came to join our diploma program at ICTP. And if I remember correctly, you came actually from a solid earth background, but then you decided to do more kind of ocean atmosphere interactions. And so Rondro was so keen and so I was also her diploma thesis tutor together with Annalisa Braco. She was so keen to study further that she also joined the our common ICTP university of Trieste and OGS PhD program. And she conducted her PhD studies also here under the supervision of also Annalisa Braco and myself, I was a tutor. And so she wrote her PhD thesis already on variability of the Indian Ocean from inter-IU to Decade Time Scales. And after this important step, she moved on. Actually, her first step was then some shorter periods, I think at Georgia Tech. You moved on to the George Mason University and Professor Shukla and Devin Strauss are here from that university and had your first research experience actually as a postdoc there. Then you moved on to Singapore and continued to research there. Then Cape Town and now you're also affiliated with the Bjergnes Center in Bergen, right? And so while at the very beginning, the focus was more on just variability of the Indian Ocean, then in later years, Rondo focused a research more on the impacts of Indian Ocean variability and what's going on in the Indian Ocean in general on the climate of Madadaska and also specifically on South Africa. And in particular, we should also mention that Rondo has made important contribution to the latest IPCC report. She was part of working group one assessing the climate in that region which is extremely vulnerable to climate change and has therefore made an important contribution to this report too. So thanks for giving us a presentation then. And it's okay, we can have it. So then we have of course Narendra Arya who actually you finished your PhD in India, right? And if I understood correctly, your first experience was more on experimental, analyzing experimental data but always focused on air pollution and composition of air and how that may impact the health of humans. But then in your post-doc period, you went to the Max Planck Institute in Mainz, right? Where you probably met Andrea Potts I guess there and who was a former staff member of ICTP and so it was then when you came more into the combining modeling with observed data and you combine these two methods to gain even more insight into this phenomenon. In fact today we had your presentation at our conference on ENSO and monsoon dynamics. It was very interesting for us to learn how actually we are used to think of how aerosols for example can influence the monsoon. This is what I've learned in the past. It was the first time I heard a detailed study on how actually monsoon dynamics can influence the composition of the atmosphere. So there's an interaction between these two phenomenon which is extremely interesting. And now you're back to India and you're doing extremely well and performing high performance computing, modeling and combining this still with observational data and experimental data. Thank you very much to be here and to give you a presentation. Hi everyone and thanks very much Fred for the very nice introduction. It's always nice to be back here at ICTP and as Fred said I started my career in climate science here at ICTP. So this is like my academic home. So for the past five years my work has been focusing on sovereign African climate and extending a bit to the eastern African climate. So in southern Africa it's home to 180 million people and the majority of the population actually heavily rely on rain-fed agriculture. So it's really important for us to understand the mechanisms that drive the climate over the area. But also on the other side it's important for us to understand both processes so that we can do some process-based analysis of the state-of-the-art models that are used for future projection. And they are heavily used over the continent, over the subcontinent for future planning. So the work I'm going to present here was done under the umbrella of future climate for Africa. And this is really a teamwork so I want to acknowledge my collaborators, Ross Blemme, Fabian Tespio and Chris Risen. What we see here it's a simple sketch of the drivers of the sovereign African climate. So the colors are the topography. As you can see that the sovereign African area it's characterized by a very complex topography. But then there are also different large-scale signals that force the climate of the subcontinent, for instance, and so. But what they have in the plot here is the subtropical high, the masculine high, and the subatlantic high pressure. We have the impacts of the warm Indian Ocean and the cold Bangalab welding system from the Atlantic. And then we have moisture transports from the Indian Ocean itself, from the tropical Pacific and from the Congo basin. And the climate of the area it's driven by a very complex mix of these processes. Now if we look at what the state-of-the-art models are doing over the area. So this figure shows the precipitation bias in the ensemble mean of the CMIP-5 models. So we have this large excess in rainfall over mainland Africa and then deficit in rainfall over Madigascar. And this precipitation bias it's not only in the CMIP-5 but it's been persisting from CMIP-3, CMIP-5, and it's still there in CMIP-6. So the question is why, what we do not understand by having such a large bias. And if I am to draw a box here over mainland Africa and another box over Madigascar and look at this precipitation, this is what we have in the CMIP-5 models. So again positive biases over southern Africa and negative over Madigascar. And it's pretty consistent in all the, in most of the couple models. And some of the models actually have more than 300% rainfall compared to the observation. So that motivates us in trying to understand what actually drives the precipitation over this area. This figure it's been shown in the activity on NSOL teleconnection this week. So this is the global rainfall during January, February, March from observation. I think we use CMAP here. And one of the particularities of the sovereign hemisphere is the presence of these sovereign convergence zone. So we have the sovereign Indian Ocean Convergence Zone over the Indian Ocean. The SACC or sovereign Atlantic Convergence Zone over the Atlantic and the SPCC over the Pacific. What's particular with the Indian Ocean that as the SACC, there is actually Madigascar on the way. So it's really in the middle of the South Indian Ocean Convergence Zone that we have Madigascar. And if we look at the moisture flux or the circulation in the southwestern Indian Ocean, then we clearly see this Easterlies from the Indian Ocean somehow blocked by Madigascar before the moisture is going inside mainland Africa. And that's because of this Mozambique Channel Trap, a trap that forms between Madigascar and mainland Africa. So the main focus of this talk is really to understand what's the role of this Mozambique Channel Trap for the climate, for the rainfall in the region. So just to explain a bit the Mozambique Channel Trap. So it's a low pressure area in the Mozambique Channel. And as I said, it's on the path of the South Indian Ocean Convergence Zone. So it could impact the southern African rainfall. Two possibilities that can trigger the formation of the MCT, or Mozambique Channel Trap. It's the dynamical adjustment of the Easterlies flowing through Madigascar. So we saw these Easterlies then blocked by the topography and then there is this low pressure on the lee side of the mountain. That's one case, but the other case is also that it's thermally forced. So forced by the thermal forcing over Madigascar. So what we did was we quickly looked at these parameters, NETA, by every scenario at all, which compares the impact of the topography by dynamically adjusting the flow or by thermally forcing the flow. So NETA, significantly less than one, means that the effect of the topography is dominant compared to the effect of the thermal forcing. And in the analysis we saw that it's 0.07. So we were quite confident that the formation of the MCT is dominated by the topography of Madigascar. So what we did is that we did some sensitivity tests with regional climate models. And we gradually reduced the topography of Madigascar in these tests and then completely removed Madigascar from the map. So what I'm showing here is the control run where we have the full topography and the flat where we flattened the topography to 300 meters. In between these we did some experiments by gradually reducing the topography by 75%, 50%, and then flattened and then completely removed. And I will go back to this no Madigascar experiment later on because what we did in this one is that we replaced Madigascar with an interpolated SST from the surrounding grid points. So that kind of creates like an artificial warm anomaly in that area. So the model we used, it's worth around a 10 kilometer horizontal resolution and we run it for 17 summers. So as I said, the physical parametrization in all the experiments, they are all the same, but we just changed the topography. So the plot in the bottom here, it gives the latitudinal profile of the topography of Madigascar in those experiments. So the dashed lines are the maximum topography and the bold lines the min topography. The first thing we got from that experiment, when we plot the 850 ectopascal wind and geopotential height, the first one, the first figure here from CFSR, which is used as our boundary condition in the experiment. So we have this beautiful MCT over here. And then in the control, we still get the MCT. But then in flat and sea, we do have these anomalous easterlies going straight from the Indian Ocean to Southern Africa. And that's expected because we removed whatever blocks for moisture. So the mechanism for that is mainly by having this flow going through a mountain. It conserves its vorticity. And then on the east side of the mountain, we have this stretching because of the conservation of vorticity and formation of the cyclonic circulation on the east side of the mountain. So just to look at how blocked and unblocked the flow is. So we use the fruit number. It's a common number that's used to characterize blocked flow. And from fruit number close to zero, it's like the flow is completely blocked. And then when it goes higher, the blocking is becoming weak. So in the control run, we have 0.15. If we keep 75% of the topography of Madagascar, then the fruit number is 0.43. And if we flat Madagascar, then it's 0.48. And this number shows that there is actually a non-linearity of the impact of this blocking on the flow because 75% of the topography is still pretty much the same. It's almost the same as the control run, but if we look at the fruit number, it's almost the same as the flattened topography here. So that's something that we need to understand is there any threshold in the blocking that gets this flow going straight to the mainland Africa? Now, if we take the MCT index as the area of range of relative vorticity over the Mozambique Channel and plot it in the experiment. So what we see here on the left side, it's the annual cycle of the MCT. The red one, it's the CFSR, which is used as a boundary condition. The magenta of the control and the green and blue are the experiments. And as we can see here that from December to April, April was the end of our run, we have this clear negative relative vorticity in the Mozambique Channel. Whereas when we remove the topography and again here by keeping 75% of the topography, it's almost the same as flattening or completely removing Madagascar from the map. So if we do the mean of the January, February, March, which is the lowest relative vorticity, this is a negative value, with respect to the mean topography, that's what we get on this figure on the right here. So the red and magenta are respectively the CFSR and control. And then we get this large jump in relative vorticity by changing slightly the topography. So what we did is we looked at the difference in mean between flat and control and C and control. And that's because we want to know what's the effect of flats. So the arrows here are the moisture flats, anomalies, and the colors are the divergence. So it's pretty clear here that there is this large easterly anomalies going from around Madagascar or in the Mozambique Channel, going into mainland Africa, and also this increase in convergence over mainland Africa here. Now, if I drew a line here between Madagascar and Mozambique and calculate how much moisture is actually going into mainland Africa in that. So in the control, it's 49 in the flat and C is 79 and 89. And again, that shows that by changing the topography, then we almost double the amount of moisture that's transported from the eastern Indian Ocean to mainland Africa. So what's the impact of that on the precipitation? This is the difference between flat and control in the precipitation, large rainfall over mainland and deficit in rainfall over Madagascar. And these harsh areas are where the signals are significant. And same for the sea experiment, it's just that the amplitude is intensified because that's quite an extreme case. So we have this dipole-like precipitation anomaly with significant increase over mainland and decrease over Madagascar. So if I go back to the interpolated SST, as I said at the very beginning, that the way we did the sea experiment is to interpolate the ocean grid points in the surrounding area to where Madagascar is. So that could create a very artificial signal in our results. So what we did is we performed another experiment hit. It's exactly the same as sea, so like no Madagascar, but we doubled the amount of heat we put where Madagascar is. So by assuming that there is a linear response, of course there are always some nonlinearities, but we assume that there is a linear response. That's what we get by having warm SST over Madagascar. We have this strong cyclonic circulation here and then less moisture and less rainfall over mainland Africa and more rainfall over Madagascar. That's to say that the signal we have here, it's actually, of course there is an impact of the heating over Madagascar, but it's not really by increasing the rainfall over mainland Africa. So one of the questions that I got when I proposed this study was that no one is going to remove Madagascar from the map, so why would you do that? That's very theoretical and I completely agree with that. It's very theoretical, but at least we have an understanding of why the topography is actually important, representation of the topography is important. So Callum Maday produced this figure. He looked at the topography profiles over Madagascar in the CMIP5 models. So this 12-degree south, it's over the northern tip of Madagascar, and going south to the end of Madagascar. The different lines are the different topography in the CMIP5 models. And as you can see, as we go southern, there is this large discrepancy in the topography in the models. And here, for instance, it's not even removing 75% of the topography. It's more than that. It's more like flattening, completely flattening Madagascar. And if we look at the difference between the low topography models and high topography models, if you look at the difference in precipitation, this is what we get, precipitation and low-level wind at 850 ectopascal, very strong easterly anomalies from the Indian Ocean and this deficit in rainfall over Madagascar and accessing rainfall over the mainland. And if I bring us back to this one, we have the bias in the CMIP5 models. We have this typo-like signal. And of course, I'm not claiming here that all the biases in the CMIP5 models are from the topography of Madagascar or the Mozambique Channel trough. But it kind of gives us a hint that varies the contribution of the topography of Madagascar through the Mozambique Channel trough in these large biases in the model. So far, I hope I've convinced you that the topography of Madagascar actually modulates the southern African rainfall through the Mozambique Channel trough and the strength of the Mozambique Channel trough actually modulates the amount of moisture that's transported from the Indian Ocean going into mainland Africa. And lastly, the difficulties that climate models have in representing the mean rainfall in the area could partly due to a misrepresentation of the topography of Madagascar and or the MCT. So that's the mean rainfall for January, February, March. The question is, does the MCT have any variability? Does it vary year to year? Does it have any inter-annual variability? So what we did is we looked at the MCT indices. Here we tried different indices. The red one, it's the one that we used in the model by just taking the relative vorticity of the Mozambique Channel area. The blue one is by taking the geopotential height and the green one is by looking by taking the first component, first UOF component of a relative vorticity. And of course all of them show like inter-annual variability. However, this discrepancy is here. We don't fully quite understand why it is in the PC one but then in the height. It's because the height is kind of sensitive to large scale signals. So if we have changed in the large scale like the mascarine height for instance then you have this peak here. So we're more confident to use the original MCT index for relative vorticity of the area. So what we did is we classify these years as weak and strong MCTs. And then look at the vorticity budget in the weak MCT years and strong MCT years. So the figure we have here on the left it's the vorticity budget of a strong MCT year. And as we can see that the tilting is dominant and that's very obvious in the strong years. And also the vertical advection is dominant in the strong MCT. If we take all the weak MCT years and of course we still have the tilting because that's from the dynamical adjustment of the flow but the vertical advection completely disappeared. Now if we do the same analysis for the specific humidity by doing a composite of strong MCT years and weak MCT years then we have this positive anomaly of the Mosaic Channel extending to the surface of the Indian Ocean and a negative anomaly over mainland Africa and the opposite for the weak MCT years. So these kind of suggest that the MCT variability can be associated with moist convection over the Mosaic Channel. And what's the impact of that on the rainfall variability? So again doing the composite analysis using the CHIPS rainfall data and SIMUP data and these signals are pretty consistent that we have this positive anomaly accessing rainfall over Madagascar and a negative anomaly over mainland Africa by having a very strong MCT and vice versa for a weak MCT. And then again looking at the anomaly cloud cover these are very consistent with our precipitation results. If we look at the moisture transport and divergence in the analysis by the strong and weak MCTs then for the strong MCTs we have this anomalous westerlies going into the Indian Ocean so it's like we have less moisture transported into mainland Africa whereas we are opposites in the weak MCT. So to summarize we know that the MCT has an internal variability and it's associated to be moist convection within the Mosaic Channel and strong MCT years tend to have excess rain over mainland Africa and deficit over Madagascar. One thing with this Mosaic Channel trap the MCT index is that now it's been used as one of the automatized evaluation metrics for southern African rainfall in the SIMUP models so hopefully this will appear soon in the ESM file too. And thanks very much. Questions, Radha? You do have questions. Thanks for the wonderful talk. I had a question about the representation of the Mozambique trough in SIMUP 5 models and I was wondering if they're able to represent the variability correctly or it's just a general bias that's in the SIMUP 5 models. So we haven't really checked the variability in the SIMUP 5 but we've looked at the main MCT and yes it's there so when you have weekend city you tend to have this large rainfall over mainland Africa in the SIMUP 5 models. Because I've also been looking at some similar things and from what I've seen it seems like in the SIMUP 5 models that the variability is completely overwhelmed by the Mozambique trough and maybe that's something we can talk about afterwards. I was wondering that giving this very important role of the topography that you have found have you had the chance to look at the ensemble of the Cordex Africa to see if you get any hint of improvement? No, the reason why we didn't really look at that it's that the boundary of the Cordex Africa it's very close to Madagascar and I wouldn't really make sense to we could look at that but I think it would be affected by that boundary really very close to Madagascar. I would be curious and you show this high mountain minus low topography case is this one-to-one with high resolution with this low resolution or do some models have a fix to pose some you call it envelope or graphy or I don't know is it a one-to-one? By doing a mean of models that have topography higher than the observation and then mean of the models having topography lower than the what's in the era or CFSR so it's not a comparison like one-by-one one-to-one model but it's like a mean of the models classified by the topography but yeah it would be interesting to look to compare with the high resolution version of one model for instance that improves yeah. Okay if there are no more questions we can thank again Rondra for this exciting presentation so good very good afternoon to all of you and thanks very much for coming to this talk so I will take you to another part of developing world it is Indian subcontinent what we call as South Asia and we will focus mainly on atmospheric traces and aerosols so earth's atmosphere mainly is composed by nitrogen, oxygen and organ about 99.9% of atmosphere but then this remaining 0.1% is really really important in the future role in the air pollution and climate change so that is how we develop a lot of interest into it but somehow a series of complex processes related to physical process atmospheric dynamics and then a lot of chemical reactions which happen in presence of sunlight that we call as photochemistry and then whatever pollution is at one place it keeps on getting transported through horizontal winds and other vertical moments so it keeps on basically all different type of process they come together and then the net result is what we see the distribution of traces and aerosols in the atmosphere so as you can see the major sources there are really diverse sources like even natural trees, forests they also emit different type of organic compounds agriculture you know a lot of ammonia and other things that are emitted from them they can also convert into particles like aerosols then biomass burning is also of different types for example they can be in the natural wildfires in the forests and we expect that in the climate warming they can increase more but there are other types also like crop residue burning done by farmers to clear their fields after they basically get their crops the type of all diverse emissions they undergo a lot of processes chemistry and transport and then what they can affect these gases and aerosols they can affect human health especially the particles which are smaller in size they can really penetrate through the lungs and they can create haze and they also affect vegetation and crop productivity so with this they are really important and we really need to know and then we need to pinpoint their sources not just at global or regional skills but to the local skills to actually inform policies like how to improve the air quality and how to mitigate their impact on the climate so in this direction space based observations did provide some good clues this is satellite data of nitrogen oxide you can see some large enhancement in the this region is called the Indo-Gangetic plane in north India and this is eastern Asia and if you can see this Indo-Gangetic plane so this is most densely populated region in the world so you can see the population distribution here so while more population they can have more emissions but at the same time a larger population is at a health greater health risk in these regions and some global model studies so basically these models take the information on emissions from different sources and they use our knowledge of physics and chemical reactions to calculate the concentration of aerosols and dress gases and then they use metrics to calculate what could be the impact on human health they find that especially human health can be more in the east and south Asian regions these numbers are basically telling in one degree by one degree are 100 kilometer by 100 kilometer how many excess premature motilities can be caused by aerosols and dress gases like ozone and in another study they use another model and they compute the ozone distribution over India and they estimate that the effect of ozone pollution on agriculture is that loss is efficient to feed about 94 million people who live below the poverty line so the impact is not only on health but is also on agricultural productivity and many of the sustainable goals of the united nations they cannot be achieved until we achieve kind of clean air and reduce the levels and another process that is becoming more important is climate warming so in Europe they found that with increase in the temperature there can be more ozone in the future one because the chemistry can be more intense with higher temperatures and another because with climate warming the forests and trees they can emit more biogenic compounds but studies in Indian subcontinent they are still lacking so we need to because Indian conditions in both chemical environment wise and also climatic condition wise they are very different and these studies not only are important for the atmospheric composition or south Asia but through monsoon the air can get uplifted vertically and then the air masses are trapped into what we call as monsoon anticyclone and then the atmospheric composition as far as the Mediterranean all are impacted Lelywild et al. wrote that monsoon has two faces like a genus head so it can inject the pollution from lower troposphere to higher altitudes but at the same time it also is a remarkably strong cleansing mechanism so it removes a lot of pollution and it creates a lot of hydroxyl radicals which help cleaning the atmosphere so with this background basically the studies and especially studies combining experimental data with models they are really important in south Asian region so we use models and we especially try to use regional scale models so the idea is that they take the transport from outside regions from global model data sets and then from within that small region we do the high resolution simulation then for we divide the region in small small boxes and we calculate the all the production loss and influx from outside and what goes outside from that box and this is done for each small small box within that region and we use a high performance computing facility at my institute in Ahmedabad and the model that I am using is called weather research and forecasting and this is very popular model it is used very widely to give you weather forecasts in US Europe and in India also but the major thing is that we are coupling it with detailed chemistry of the atmosphere and then we are running it a high resolution generally global models their resolution is about 100 kilometer by 100 kilometer but to go really up to the local scales we are doing simulations at 12 kilometer by 12 kilometer so and then we have to put some simplifications of the chemistry and then we have to because we have to India is really large region so we have to cover entire India so to save the computations we have to use simple chemistry and I will show what those simplifications how they affect the model performance most importantly since these models were developed for different parts of the world and when we are applying to developing Indian regions we have to do rigorous evaluation of the model before we conclude much on health impacts or policy formulation we need to have a rigorous evaluation which I think was the biggest challenge because data sets are very scarce in developing Indian regions so I will show a lot of our efforts that we have made in this direction and besides our own modeling we also use reanalysis data sets like data sets so similar to that here we are using some chemical reanalysis so like if we have satellite data set that can be assimilated with model so these type of data we can also use for India so this is our domain of the model this is covering almost entire India some extra simulation we are doing to cover even bigger regions and what information goes into the model is basically the emissions of all different gases and aerosols from different sources most importantly anthropogenic sources like fossil fuel burning then biomass burning is coming biogenic emissions are calculated online because as I said earlier that emissions from trees vegetation they depend upon the temperature so those are calculated using the weather conditions and then we compare we got several stations although they are not kind of covering a lot at high resolution in India but we can make some comparison with the observations and there is a lot of uncertainty in the emission data itself so we use three different emission inventories for India and we found that output can really change if we use one emission inventory as compared to other so this is an important challenge for developing regions that emission input they are not very fixed they are not very accurate so substantial uncertainty into our conclusions can pop up from that and not just the emission inventories it is also important how much simplification is done into writing the chemical mechanism of the model so if we use one chemistry mechanism if we focus just on this blue line and this pink line so when we basically compare we keep everything same just we keep more details of the chemistry then also model outputs are changed a lot so by comparing lot of data sets we came up with the configuration with which we can really reproduce in a better way the observations that we had in India so for example here I am comparing our simulation with a ship cruise based experiment in the Bay of Bengal so you can see that these black dots are basically the observations collected by during the ship and this blue line continuous line is the our simulation so there was some difference in the levels but if you see the variability so model is capturing that same way we are comparing ozone here we did additional simulation in which we turned off all the emissions happening in India and then this line become this dotted almost flat type of line it tells that the variations that ship based experiment list are seeing the Bay of Bengal to great extent they were controlled by the outflow from the Indian processes like so this is we are putting into a simpler form here that air is entering from this Arabian sea and while leaving India towards reaching the ship based instrument it is the photo chemistry is happening exactly in this part of the region so this type of information we can get from the when we combine the observations and model so this was ship data now let's see how the data of model compares with the aircraft based experiment so aircraft based data was collected from by putting instruments into Lufthansa airline which used to fly between Frankfurt and Chennai so when the aircraft takes off from Chennai they get kind of one profile up to about 200 hectopascal that is the cruising altitude of long long flights and then when it comes back to Chennai and lands then they get another profile so taking this type of experimental data we compare with our model and this is basically distribution of water vapor so when models capture this type of vertical distribution of water vapor we get confidence that they are getting the monsoonal convection right so we go again to the satellite based observations they can also help filling the gap of experimental data and what interesting we find is the left this panel is basically our simulation and these are data from three different satellites OMI is NASA's ocean monitoring instrument GOM2 and CIMHCR from European Space Agency so here we are comparing one particular species called formaldehyde the scientific significance is that most of the hydrocarbons when they undergo different processes in the atmosphere they are converted into formaldehyde so if formaldehyde is high that means that that particular region is experiencing intense photochemistry and the technical importance of this species is that it has some spectral signatures so it can be measured from satellites other species are also important but they cannot be measured through satellites so with this we compared our model output with the different satellites so what we noticed is that our model predicts other than the endogangetic plane besides that it is also showing a lot of hot spots in the regions which are not that much dominated by anthropogenic activities and these regions are basically dominated more by forest area and natural emissions so that was one important finding that because in the future when climate warming will happen probably this type of process will become more important the natural biogenic emissions they can be more enhanced the other important point that we conclude by comparing these satellites is this that there is some differences between the model and satellite but there are also substantial differences if you compare one satellite with another satellite so this means that we need to establish experimental network of stations within India so that we can validate both satellites and then we can pinpoint the uncertainties in the model so we applied this model to a particular problem which was of interest for wider public interest also and for scientific interest also that in India every every towards winter we see a lot of hedge covering entire endogangetic plane this is really widespread so these are numbers are called fine particulate matter there as I was saying that small particles they affect health more strongly so this is the distribution of those and to make sense of these numbers generally generally the level should be less than 25 generally the level should be less than 25 to do not affect our health too much but you can see that most of the regions it is really widespread and levels are exceeding by order of magnitude so the impact will be really remarkable on health and it happens typically towards every post monsoon and winter that most of them are this endogangetic plane that experiences really really high levels of the particulate matter so we did the simulations these are our simulation plots and we try to compare it with some of the stations located in the gangetic plane and although model does not perform perfectly but there were some agreements between the time variations that we saw in the observations when we compared the model output so what we did we removed the biomass burning in the model and calculated what is the difference that is coming because of the biomass burning emissions so we find that in the Punjab Haryana regions where most of the crop residue burning happens so their strong effects were there from the fires and they were also persisting and spreading around during post monsoon like November but when we have the maximum pollution loading that is in winters then the or gangetic plane the effect of biomass burning was not there so it clearly tells that biomass burning effects they are only important in the post monsoon and when the PM 2.5 was the highest then fire effects were really not important at least for the gangetic plane some biomass burning is still visible over the Himalaya but this is related to the forest fires and these emissions they do not sink down and they are more confined within the Himalaya so with this we try to kind of communicate what are the important sources and what are not and why this type of enhancement are suddenly appearing you can see that towards winter temperature is really really very low and boundary layer depth that tells us how vertically the dilution of pollution can happen that is very very shallow so all the aerosols they are confined within a very small volume near the surface so at that time if sources like other anthropogenic sources they are high so the pollution problem is really really very intense then in the previous talk also there was a lot of highlight about representing the topography of complex regions so in India also we have a really highly complex terrain that is Himalaya and it is I was mentioning I was mentioning that it was just above this Indo-Gangetic Plain and kind of which is also affected by a lot of biomass burning fires this region is highly complex in terms of topography you can have a look that this region you see a lot of houses this is basically what you call Indo-Gangetic Plain and for this I was mentioning that this is most densely populated part in the world but if you see that very many these mountains are located so within this region if you want to do modeling so they often models do not resolve a highly basically plain regions placed very close to highly complex terrain of the Himalaya so what we do we go to very high resolutions so this star is basically this location this is in the central Himalaya and this is basically an astronomical observatory I was basically doing my PSD from this institution it is called so I basically downscaled the model to really really high resolutions at one kilometer by one kilometer and we tried to see if the models are becoming any better so this is common problem of the global climate models that in mountains they kind of overestimate temperature because they do not represent the mountain tops very well so when we go to from coarse resolution to high resolutions we really improve the resolutions of temperature for the central Himalaya so that was kind of bit promising but there was some discrepancies that in the local scale atmospheric dynamics model is not getting the winds direction properly so this is the observation observation says that most of winds are coming from east but at coarse resolution this component was missing at high resolution it appeared but it also appeared with some discrepancies right so the result is that we could better reproduce the temperature variations in Himalaya but discrepancies in winds are still substantial and there are local scale processes so probably it will still be challenged to even at such high resolution it will be a challenge to reproduce this type of environments so when the human intelligence was kind of giving limited performance we tried artificial intelligence also for Himalayan environments so this is one exercise that we have recently initiated we use machine learning algorithms and first we try to train the machine learning algorithm with 2-3 years of observations and then we try to test whether it can predict the remaining observation independent data sets but to really exploit or really test the ability of machine learning we took long term data from the chemical reanalysis model data like temperature humidity, boundary layer height, winds and some precursors of ozone like carbon monoxide, nitrogen oxide etc. and when we put all those arrays into the machine learning algorithm then model performance was really good so we can say that machine learning based models they can really perform well and they can help in the prediction of ozone and probably this type of exercise we can extend for aerosols and we can try to explore more potential of machine learning to complement the conventional regional modeling that we are doing and we also analyze this type of long term data and find one important interesting result that while sulphur dioxide that is a tracer of emissions from for energy generation like power plants etc. coal burning etc. so earlier the trains were very fast or India but in recently they are slightly select they are showing some signatures of stabilization and trains are becoming smaller now so this type of information can also be generated when we combine the satellite observations with models so to summarize what all I discussed is that several important features that we saw in ground based, ship based aircraft and satellite data sets or Indian subcontinent we could reproduce many of them with good ability of course there are some more biases and we need to work them out and we use the model to discuss an important problem of widespread winter time pollution across Indo-Gangetic plane and we found that biomass burning does not have the strong role at least in the winter time its role was important but only up to the post monsoon using our simulations we could also predict that some natural sources are also important and their role can be enhanced in the case of climate warming and it can be more important in the future as we see a slow down in anthropogenic tracers like sulphur dioxide and we could resolve the Himalayan topography better as compared to previous studies at high resolutions and improve the temperature variations but still there are substantial discrepancies in the simulating the local scale dynamics of the Himalaya so it will take more efforts to really study the transport of pollution from gangetic plane to the Himalaya and see what could be its impact on climate and we did some initial exercises using machine learning methods and we find that they can help especially in the regions where conventional models are not performing very well and especially if we can have some long term experimental data sets to train the machine learning algorithms so I also like to thank many experimentalist friends and collaborators who have been helping throughout the work and thank you for your attention Thanks for the questions Very nice lecture Dr. Reza the role of biomass burning we really hear different claims about how bad it is and how important it is obviously newspaper reports are filled during the winter season the whole Indo-Gangetic plane gets kind of quite affected so what is the evidence based on which you are saying that it is not a serious problem during the winter because the impacts are there so what is the main reason for that winter problem Right as you rightly pointed out it was an impression that biomass burning especially for field clearing that is kind of responsible for this widespread haze but these type of fires they can be detected from satellites so the emissions from fire that we included in the model they are based on satellite based observations of fires so once we included in the model then our model was reproducing the day to day changes in the aerosols over this region and then when we remove that fire emissions from the model then we see that there was not much effect on the Gangetic plane's air quality so fire's role was important and it is really really important at least during post monsoon when the crop residue burning is happening but after some time these type of aerosols their lifetime is two weeks or so so after that their effect is gone so then the major thing is happening that it is two cold conditions across the Gangetic plane and whatever remaining anthropogenic aerosols are there from non-fire sources they are trapped into very shallow stagnant atmospheric conditions so to counter this type of effect produced by sharp change in mutuality we have to rigorously bring down the other sources also just expand on the non-fire related anthropogenic sources what are they just can you mention a few of them right so towards winter some the other sources like fossil fuel burning industrial emissions they are all time there but towards winter the additional other sources come for heating purposes heating purposes and residential purposes so they are also getting enhanced I would be interested if they are in this in the footholds in this pollution area if there if you know if they are strong dynamically induced inter-annual variations of this pollution or if it is essentially every year almost the same okay so every year this is a typical cycle that towards post monsoon and especially winters the haze is most most pronounced and it really affects a lot of flights and they have to get cancelled due to poor visibility so this is a typical phenomena and a lot of research is happening on this and but within the one season to other season there is a lot of change in the air quality from one year to the other you just expect the same kind of more or less the same kind of suggest that important seasonal changes they have somewhat control on it and some new studies are coming up and they are telling that there are some chemical environments which allow more condensation and aerosol growth in these environments so those angles are also being investigated right now so there are some puzzles still in this problem the general variability of circulation from year to year can be a source because the inversion layer once it so that has nothing to do that has to do with the sort of climate variables it can it is playing a role is that correct there is a significant some contribution also of just the weather and climate state itself right so in the years which are kind of colder than other years the effect can be more there are no further questions or comments what is the next item on the refreshments on the terrace is there is no other photo right the photos we have done with the photos ok ok ok ok let's thank again then both and we have a refreshments on the terrace