 The charts. This was that system there that developed and moved to the south across the peninsula. And she brought quite a bit of precipitation on Mandi to us. Isn't there this interesting comma feature that sits around all day there just off the East Coast of Scotland? I don't know. I don't have a chance to look into it very much, but it just sits there and spirals around, by the way. So I'll skip over this, actually, just to be super quick so I don't want to hold you. Felly, mae'r ffordd yw'r ffordd yw'r rhaid i'r hollwch, sy'n gallu bwysig i'r regionau sy'n cael ei wneud ar y 25th, ac mae'r ffordd yn cael ei wneud ar yr hollwch. Felly, yw'r ffordd yw'r hollwch, ac mae'r ffordd yn cael ei wneud ar y 4th, oherwydd, mae'n cael ei wneud ar y month yma o'n Llyfrgell. Mae'n cael ei wneud ar y month, Ac so we had a couple of big events of about 20 millimetres on the 22nd we had 12 millimetres. That event actually only brought two and a half millimeters to the traditional station. So that it shows you here there was actually quite a bit more precipitation up in the castle. So we're on 46 millimeters for the month so just in case you're interested that said about the 25th percentile so the average is I think on the orders 60-65 millimeters in terms of the climate. Ieithi'r rai, 64 miliwn. Rhaid i'r rai lywodol yn gwneud, bo'r grannu clifennu yn y meddwl i'r oed yn ddechrau'r gweithwyr a'r gweithwyr rhaid i'r gweithwyr. Wrth gwrs, yna hynny, y cyffredin yna yno, yn gweithwyr 64 miliwn. Felly, yna yna hefyd yn fyrwyr i'r llun o'r fath, mae'r ddechrau'r gweithwyr yn ddwy'r ddwynyr, sydd yn ddim yn dweud, yn ddwy'r 44 miliwn yn 2012. The next rise was actually June in 2006, where we had no precipitation at all, just one millimeter. So then just super quickly, because I don't want to delay you any further when you restart. This is then the EPS gram for this week. So just to remind you, I forgot to mention the dark blue line is the high resolution deterministic system. So this is the grid box nearest to Riesta, adjusted to station height. And all these bars and whiskers are basically representing the other 51 EPS members at a lower resolution to give you an idea. They have different initial conditions to give you an idea of the uncertainty. And you can see a lot of these are kind of poking up and showing some signs of precipitation we'll talk about of sherry periods. And increased winds on kind of Saturday through Sunday, which we'll have a look at the wind direction in a moment, but it's likely to indicate borah conditions. And normally this is an underestimate, of course, because even at the very fine resolutions of ECMWF, you're not resolving the flow through the gap here. There are three borah regions, one here in Riesta, one down near the island of Caso. So last week they actually had a much stronger borah down in Caso, and then down near the island of Pag and Crocia. So you see these three distinct regions where you get this accelerated flow. So if you look at the Met Office forecast, some wind is back. So those of you that are here from the UK, unfortunately, while you've been here, of course, it's been this massive blocking high over the UK. They've had 28 to 30 degrees in London all week, and it's been pretty nice. But that's actually that development now is going to move off to the northeast. Unfortunately, just in time, some of you if you're going to be here next week anyway. So it seems that you're just too worried, but that development will basically move off to the northeast over the weekend. But if you notice here, we've got a sign of another convection line again with the north leaf flow giving us unstable conditions. We can pick those up on the radius on, and this is going to be moving down through Croatia over the weekend. So on the EPS grant, a lot of the members didn't have any precipitation for Trieste, but we'll see at the moment in the map that essentially there's a lot of convection just up over the castle. And it's very likely that some of that could actually, some of that could hit us as it moves down. So we go through Saturday, you see that development moving down. And then you'll see that there starts to be a low development over the Atlantic, which will bring disturbed change conditions to the southwest of the UK. So we'll find there are two periods during the week. So if we actually go to the maps of the WF forecast and wind back to this weekend, as I was just zooming forwards through Saturday, you'll see this is basically, we've got the pressure lines on here. These are the precipitation from the deterministic run. And you'll see this event, I've winded one right through. But if we go through Saturday, you can see those shells right through Croatia, just slightly off. And then they move to the southeast. And then later on in the week, I'm going to zoom right forwards because I want to do this super quickly. We find that Tuesdays through Wednesday as well. We've got a lot of precipitation over the Alps and showers right across the region again. So it looks like especially midweek, of course, that's a long lead time, but it could be very showery again Wednesday, Thursday, and it's likely that we might get hit by a few of those thunderstorms. So it's looking mixed next week. It's some nice days, but towards the end of the week getting showery again, I'm afraid. So it's a case of enjoying the weather, but make sure you've got a little mini umbrella with you if you're out and about, especially in town. The temperature range, if we skip back to the EPS gram. Now, you'll notice this is very typical for the, this is this bottom chart here. Now, basically the deterministic is on the kind of mid-20s. You'll notice the EPS often has the peak temperatures much warmer. Okay. And that's due to the fact that complex topography, even if you do a station height adjustment, it doesn't account for the fact that if you're next to the sea or up in the castle, there's a big difference in the actual, should we say, the temperatures. I mean, in the winter sometimes it can be seven degrees difference between Oppochina and here, even though it's only 150 meters difference in height. So it's likely to be in the upper 20s. We're talking about 26, 27, 28. Okay. Thank you. Thank you. Sorry, I think so. Okay. So, you enter on your variability. So, one of the term, if you talk to carbon cycle people, often you hear is a so-called CO2 growth rate. It's nothing but the DCO2DT time derivative of answer CO2. Mononoa CO2 data is a wonder. It tells many stories. So we can, by eye, we see the increase in seasonal cycle. But if you look at it a little bit closely, you see their variabilities under annual time scales. If you do very simple analysis, you will actually pick that out. So one thing you can do is really just to take the time derivative of this time series. If there is a linear trend, then it will become a constant. That will be the mean growth rate of CO2, which now it's like 2 to 3 ppm per year. If you recall, 1 ppm is about 2 gigaton. So 2 to 3 ppm, that's like 405 gigaton. So that's half of the fossil fuel emissions. It becomes, I would say, the growth rate. But when you take a time derivative, the dominant signal is actually the seasonal cycle, which is very large. But if you do a smoothen such as a simple 12-point rounding mean for monthly data, you end up with this curve, which shows actually the dominant signal is the inter-annual variability. So Dave Keeling and colleagues had noticed correlation between CO2 concentration and El Niño very early on in the 1970s. Even though they were not doing DCO2DT amazingly, so they just look at the CO2 concentration itself and saw the correlation. I was still in the case who said, hey, why don't I just do this time derivative? So this showed up very clearly. During El Niño, for example, 1997, 98 El Niño, you see the atmospheric CO2 growth rate is 7 ppm per year. So that's like twice as large as normal. And in fact, so much so that for a short period of time, it completely swamped the fossil fuel emission rate. So it's a larger fossil fuel emission, meaning that land and ocean together, now it's not a carbon sink anymore. For that brief period, the carbon sink completely disappeared. The land that released so much carbon actually outcompeted the ocean carbon sink and more, so you actually get a net atmospheric CO2 increase above the fossil fuel emission rate. So this correlation, you do analysis, for example, with the Southern Oscillation Index, you find that there is a delay between CO2 growth rate and Southern Oscillation Index with a max at about 5 to 6 months. The max correlation is 0.6. So this is actually quite amazing. If you think the carbon cycle is responding to El Niño, there are a lot of processes in between. If you take global precipitation, you would not get such a high correlation with El Niño. So this is quite remarkable, and for 20 or 30 years, carbon cycle scientists try to explain this, understand this. So now we have a relatively clear picture of what's going on. You look at, for example, a mechanistic model like Vegas plus the ocean side with the GFDL ocean carbon cycle model, shows that, first, in the Niño region, of course, that's where things are going. So you do have carbon flux anomalies, which in part has to do with upwelling temperature dependence of solubility of CO2, and in part has to do with biological activity change. But the bigger story turned out to be on land, especially in the tropical land regions, Amazon, maritime continent. Now we have convergence between mechanistic model simulations that simulates climate impact to the carbon cycle, and the atmosphere inversion I mentioned earlier that uses CO2 observations and atmosphere transport to infer, to calculate backward, where the source and sinks are at very large scale. So you see, again, the very large CO2 release to your Niño-Niño from places like the Amazon. And mechanistically, it actually involves a lot of interesting processes. So in this work that called this in a conspiracy theory, conspiracy in the sense that why carbon cycle response is a way downstream besides such strong correlation with the Niño. So there are several steps. One is during the Niño, sub-tropical land tends to have drought. So Amazon rainfall is suppressed. So it's a very interesting climatological phenomenon. During the Niño, Amazon rainfall is suppressed. Indonesia rainfall is suppressed. Africa, OK, here and there are different. And these regions, because they have so big carbon floxies, they're so sensitive to the tickling in the pressure and temperature change. So they all respond in the same direction because they have drought. And then come down to plant physiology. For example, when you have a drought over the Amazon, you would grow not as well. So you would take up less CO2. And on the other hand, in the tropics, if you have a drought, you also have warmer temperature. This is very different from high latitude where temperature precipitation go together here that go in the opposite direction in the convectively dominated regions. So it's purely climatological but placed into the physiological response such that in warmer temperature, decomposition increases. On the one hand, you take up less CO2. On the other hand, you respire more CO2 because it's warmer. So the geographical coherence as well as the climate plant physiology interaction leads to a very robust response of carbon cycle as shown in the CO2 growth rate to interact with your variability. And another story of the interactivity is actually fire. Fosan bagen diwrn 1979 El Niño. There was major fire in Indonesia. Every time El Niño there was fire there. But that year was very strong. They drained peatland in that region. Sometimes they drained for palm oil and so on. And released a huge amount, contributed a huge amount of the carbon release, going in the same direction as respiration. Direct combustion of biomass, not a slow decomposition. So we can look at such things both from satellite observations as well as mechanism model and understand their impact. So some of us would like to say CO2 is a climate index, is a very good climate index on these different timescales. And now that we see seasonal timescale, annual timescale, and even longer, carbon cycle really responds to climate. So that offers a hope to predict the ecosystem of carbon cycle on the inter-annual timescales. So this is some of the exercises we did. There's a high kind of handcast experiments. Imagine if Dave Keeling is live, I can tell you what amount of CO2 is going to be like next year. So now show one or two paleo climate examples. So this one I showed earlier. So Michaela was asking the proxy data CO2 reconstruction. So not only that we have CO2 reconstruction, we have all other kind of evidence, for example temperature and sea surface temperature as well as a galatiation. So like the most recent, we are in a great ice age. So during the glacial periods, the ice sheet goes to 45 degree while other cold periods where CO2 is very low, it goes even further towards equator. And of course if you get to here, you have some snowball or incidents. So this period here, the carboniferous period when the Pangea was, had all the land masses together corresponding to this period there. CO2 was relatively high, so we had a lot of warm, humid places where a lot of cold got buried. So one very interesting thing about carbon-climmy interaction is on this kind of timescale, there may be a slow interaction between climate and carbon cycle. It's not just a, only like an act and response. It may well be a truly dynamical interaction. For example, imagine a warm period you have a lot of carbon start to grow and a small fraction of that gets buried in for example wetland. So slowly I'm sure CO2 is taken out of the system. Not just I'm sure CO2, remember that's a small quantity. That's quick ocean CO2 just comes up very quickly, to compensate for that. You need to take a lot of CO2 out of the system to be buried in order to really draw down I'm sure CO2. So there might be this kind of processes with different timescales in it that, you know, at least partly contribute to this kind of variability. You know, this is not cannot be explained by Milankowicz. Right. We don't truly have explanation for this variability in the sense that we can fully explain the dynamics by the data indicates interactions. So in fact there is a thing called the CO2 theory of climate change that claims most of the climate variations throughout the earth's history involves CO2 and possibly methane sometimes starting from the Fendt-Yang-Sang paradox. So I have personally quite a little while ago quite involved in the glacial cycle problem. Mainly looking at first start trying to explain the CO2 problem. So that's you know, the modern time we have the missing carbon sang problem. Go back in time you have this glacial CO2 problem. Now we have so much data on this time scale. We can really fully explain them. And so I had my own pity theory that called the glacial very hypothesis that tries to add a piece to the glacial interglacial cycles which involves actually ice sheet carbon cycle climbing interaction not trying to explain the whole thing by just to throw in something that might be of interest for inspiration. So the idea is this. You think about today's Canada that whole boreal area it was covered under the northern tide ice sheet. Today it's boreal forest a lot of carbon we just look at the climatological distribution of carbon right? And you ask the question where does that go when the next ice age comes. So the traditional view is it's bare rock under the glaciers. So this hypothesis I just said no all that carbon at least a part of that carbon probably gets buried under the ice as slow cover this areas permanently. So later so essentially you lock all this organic carbon into a freezer. So later as ice sheet grows very large ice sheet starts to move that might push out the organic carbon and would decompose release CO2. If the release of the CO2 is fast enough it might even trigger a degallatiation even if there is no Milankovic. So that provides a internal dynamics a mechanism that could generate quasi 100,000 year cycle as simulating this simple model but very dynamic so there is no Milankovic force applied here is just internally generated it produced by itself about a 90,000 year timescale. That of course is not to say Milankovic is not part of the story. They still be the main story but just to say we have problems for example the degallatiation before the emion called termination 2 we have a causality problem we have many other problems we can't explain. So interactive mechanisms like it might actually play a role even in terms of triggering degallatiation. So back to modern carbon problem namely the so-called missing carbon sin problem. If we so the items for CO2 measurement is considered very accurate. We have all the stations small and all and so on. So we get that. Then we can estimate the fuel. We can estimate the land use deforestation flux. And these of course go into this carbon sinks ocean land and then the rest remainder is in the atmosphere. So percentage wise about half is left in the atmosphere. A quarter goes to the ocean and a quarter goes to the land. So the atmosphere scientists and ocean scientists have done a great job of quantifying these numbers. So they give a certain range here. They're pretty good. Very narrow. But the terrestrial scientists have not been able to come up with a number that is as good with small uncertainty. So we end up using a mass balance equation like this to say hey we know this we know this we know this. So land has to take up this much carbon. But we don't exactly where they went. Do they go into the for example reforestation in China or the agricultural green revolution US no till soil or different other places. So we don't know. So that's why it was called missing carbon sink. And really should be called residual carbon sink. And now there's another new term called unaccounted full carbon sink. So that's still a major issue I'll try to chase after. So as I touched upon earlier understanding these things and how predict how they are going to evolve in the future will determine what the answer to concentration is going to be like in the future. So that brings us to this great grand coupled Earth system modelling prediction where you would fully involve all these components in an interactive way. So traditionally what we do is give me a CO2 scenario double CO2. I'm just going to simulate what is the temperature change climate sensitivity and what's the precipitate change that is a climate system. Now we ask the question Okay, give me a fossil fuel emission scenario. Not in terms of PPM in time but give me gigaton carbon per year in time. We're going to try to predict this and this together. So that task was taken on about 10 years ago by a group of models in a project called the SAFE for MIP model. I'm going to go to more detail describing but to show quickly a result you can do this in a so-called coupled and coupled case. You consider climate carbon feedback. This is purely climate carbon feedback. What does it take? For example, I'll point out one process if the atmosphere if the climate gets warmer. So the solubility of CO2 in ocean water would decrease. Okay, colder water contains more CO2. So it would decrease. So that means the ability of ocean to absorb carbon would decrease. Similarly on land, if it's warmer the decomposition will be higher as we talked about in the previous hour. So these are positive feedbacks. So you include that kind of positive feedbacks in the simulation. You get results like this. You get about 100 ppm higher CO2 when you predict to 2100. And temperature like half to one degrees warmer. So that is of course a lot. A big difference. So this is that first group of models that are involved. In fact, the first such positive feedback was pointed out by a separate modelling down at the, with the Hallysinner climate model by Peter Cox and colleagues. So when all the models got put together what we find is there's a large range of prediction. This is nowhere smaller than for example the IPCC's different scenarios. And in particular the largest uncertainty comes from the land carbon cycle. So much so that the models are going in a different direction. This is a historical simulation to present and into future. But present pretty much everybody simulates a carbon sink which is supposedly the missing explains the missing carbon sink. But as you go into future for example the Hallysinner model which saw the alarm before this comparison show that the terrestrial carbon pool changed from a sink to a source. So it will be emitting 6 gigatons of carbon remember carbon fossil fuel emission is 10. At the time of simulation was 7. Well another model continues to take up carbon to 10 gigatons. So there's a huge variability. So that at the time caused a lot of interest. So IPCC immediately included in the 2007 report and then in the next report this is a so called Earth's Submodeling Framework. So another thing came out of Peter Coxscrope is a prediction that the model prediction that Amazon might actually die back. The whole Amazon would got killed in that model simulation which contributed to the carbon release. Because that's also a dynamic of vegetation model. So there's a later study looking at what causes Amazon change turn out Atlantic ST gradient that seems to be pretty robust in the climate model predictions the impart responsible for changing the monsoon ITCC movement and later analysis shows that Amazon probably would not die back the whole Amazon got killed but it's the southern Amazon that is in potential danger so as the monsoon moves northward climatologically so you become a little bit drier there's a dry season so dry season becomes longer of course this is also where deforestation is ongoing so the two things together might still pose a major threat to that region. So now let me put a summary side there to say that biosphere, climate interaction now we come to thinking in two aspects one is biogeochemical feedbacks such as the CO2 feedback but you can also think of other VOCs and so on the other one is biophysical feedbacks albedo, evaporation and roughness lens so this really modifies the three major components of the surface fluxes so a show a couple modeling example that looked at really just the terrestrial interaction and the biophysical aspects with the physical climate system so one thing is vegetation is really a is really a product of climate you look at observed vegetation index, normalized vegetation index and the satellite based with precipitation distribution you find a very nice correlation and that is non-linear saturation behavior in the simplest dynamic of vegetation model you can just put in a relationship like this here is my vegetation model and then try to convert that into albedo change which you can get out of another satellite correlation analysis and in fact this is almost what we did a few years back this work look at the Sahel drought problem you know the 70s 80s drought was probably one of the largest climate anomalies in the 20th century this prompted the UN to start the convention on combating desertification but for a long time we are not able to explain this change which is so large so in this modeling work first SST was put in as a driver so SST gradient caused the monsoon gradient a change shift in the monsoon so it caused the drought but one important thing to notice there was this SST simulated amplitude of change is so much smaller than the observed magnitude so then soil moisture feedback was added then dynamic vegetation feedback was added so go to amplitude after you add all these processes to more or less explain the observed change so this land feedback vegetation fact turned out actually had a long history earlier and a little bit later Charney made it very well known what he called overgrazing causes albedo change and then that induces a cooling anomaly over the atmosphere above this region that corresponds to a subsidence so he actually derived analytical equation to that effect so this Charney hypothesis is basically energy albedo feedback and so this kind of feedback was in the simulation I just showed and however most models for example in the C20C multi model comparison project that really look at this problem practically all models can simulate this drought when forced with observed ST there has been quite some work looking at ST from Atlantic Ocean versus Indian Ocean Pacific Ocean and so on so including Michaela's a lot of work and commentaries on this problem and later there's our aerosol all this is coming to play which I won't be able to get into here but one interesting thing is if you actually look at the amplitude of this model simulations they are all too small compared to observation even these two Google AI models including the model I just showed earlier actually does not really get this amplitude right not enough so perhaps Charney's earlier idea some human impact is still there so the model we're talking about is still natural vegetation feedback not additional human overgrazing in there so it's still an unsolved problem so Yongkang, Shu, Leide multi model comparison more recently so there's some more update on this problem so it looks like I actually have 10 minutes left so now we ask the question which a lot of people are concerned about is what's going to happen to the desert in particular the Sahel problem is it going to be a drought there's actually a somewhat of grinning trend you know after the 70s drought that you can see again with NDV and so on so this gets to the problem of drawing of the subtropics or the intensification or expansion of the highly sale I know dynamically there's a lot of debate and so on which I'm not an expert but here we look at I show a some results point out the downstream processes you know when you do I'm sure dynamics precipitation is like your final destination I predict the precipitation is going to change like this when you stop there but from impact point of view we care about what's going to happen to soil and my crops and my ecosystems so if you do a modeling like that you actually drive these dynamic vegetation models land surface models say semi IPCC models predicted precipitation change what you see is this so the places with precipitation reduction mostly subtropical regions expand when you look at soil moisture especially when you look at vegetation for example here over western Amazon you actually have slightly increased precipitation but vegetation just decreases so as you look downstream the impact of global warming becomes worse and worse that is probably a big thing and for the increasing drying in soil moisture yes and then there's vegetation dynamic vegetation process here as well current work emphasizing the that while you have a change in the productivity you also have a change in efficiency and the storm model opening and all that right so I don't think that the best assessments look like that Jack Sheff is going to present next week some of his work and calculations about how to think about drought in various ways that include that coverage whether it's meteorological drought as in precipitation hydrological drought as in soil moisture runoff and agricultural drought so they all give very different answers so I think we'll get back to this next week but I think it's a little bit more I would venture to call this ecological drought so if you again put in dynamic vegetation in a model like that try to predict the future what you find is this is an observation and simulation if you include dynamic vegetation well let me rewind so because of this kind of drying trend so the desert actually will expand from present day 25 million km2 not present sorry present days here to you know a few million km2 larger but when you put dynamic vegetation you have drastically more drying desertification certification predicted a simple model like this so of course it's just a model and perhaps we can get some one of things comes out of all these land answering interaction studies Sahel seems to be a very sensitive region this was in fact noted in Chinese on GCM simulation earlier on and compared to you know the Middle East and other places so this is saying in the paleo a record so this is the middle of the Sahara so you know archaeologists have found these rock paintings that depict dears and zebras and so on grass and so on so apparently it was a much wetter place so Martin Clawson and colleagues at Max Plankey Institute did this simulation that showed after the Holocene you know the monsoon all pretty much retreated but their simulation showed okay the solar forson is relatively gradual but there is paleo evidence for example the dust collected off the coast of Africa in the Atlantic Ocean show that very rapid increase in dust accumulation so their model is actually able to simulate this kind of rapid change after they include dynamic vegetation feedback otherwise they don't get it so both Martin's group and my group have looked at the dynamics of this kind of problem quite a bit for example the using the Klimber model they showed that there is a dependence on basic state so the present day possibly it can have a multiple equilibrium a wet state and dry state while during the Holocene while the sun is up there you may just have one single equilibrium state so you could look at this kind of problem with relatively simple equation to a degree such that you can use a mathematical to solve it for example one of the interesting thing coming out of this is if you have a nonlinear system you find a equilibrium here which could be let's say this trough or this trough here but it's nonlinear then if you add a climate variability to it so make this and go up back and forth your final average equilibrium state is different is shifted from your state without climate variability so that can be demonstrating actual model simulation for example here you start for Africa you start from old forest you find the interactive model you find it settles to a still forest lake state and if you start from a desert lake state you start from zero vegetation you end up with a desert lake state so there's multiple equilibrium here but if you introduce variability you find these two states start to converge towards each other when your variability is so large and that kind of wipes out your multiple equilibrium states finally I want to show a study that is not yet published so led by my colleague Eugenia Cone who happened to be a Chinese student who has always had this love the Chinese idea so we worked out this together and in fact so Fred Cujarski and I put together the speedy model with the biggest dynamic of each model a few years back was actually then this model was perfect for studying what might happen in the future in what you might call renewable energy or geoengineering that kind of setting there has been quite some effort including the actual project called desert tech I don't know if anybody heard of it and that actually now is aborted but seriously people talk about putting wind farms and solar farms in the Sahara to generate enough energy to supply the whole Europe and North Africa energy needs in the future you can do that kind of back envelope calculation there is more than enough sunlight there to do that so then the question for a climate scientist would be if we do that what would be the impact on climate we have seen this region how sensitive it is to surface condition so in this modelling work we basically put solar farms and wind farms over the Sahara and assuming that there is a little population despite of this and that concerns maybe people might just do it and so here is really the slide that summarizes the result coming out of this speedy Vegas model first you see that both wind farm and solar farm alone will cause warming near the surface and reduce an increase in precipitation if you put them together the effects even stronger what is surprising to still the magnitude of change precipitation more than doubled so especially south in the southern part in the Sahel so this really dynamically has to do with the modification of the local highly sail here in this region there is albedo yes, solar changes albedo and wind farm changes the roughness lens which causes low level convergence increase a low level convergence which is actually sometimes called the soot mechanism yogesh soot in front of Goddard did some earlier GCEMs simulation there so it's both energy and momentum and modification that led to this very large change we see so of course the consequence of this is very important if you we have seen geoengineer and some of them might have adverse effect and this one seems to at least by this criteria seems to be a good thing anyway so that concludes my lecture thank you very much