 So, good morning, everybody. You sleep well? Nice sleep? Okay. And this lecture is about calderas, okay? So we are going to see the structure, the evolution and the arrest, as well as the magma transfer processes within calderas. So I will just start with a quite provocative slide. This is a Google Earth view of the Neapolitan area. And you see this dark spot here. You know what is it? It's Vesuvio, right? And you know Vesuvio is famous for its, mostly famous for its Pompey eruption 2,000 years ago. And Pompey is somewhere here. Of course Pompey was there 2,000 years ago. And today there is no Pompey, but we have Naples today, which is on the opposite side of the volcano. And of course there is a significant treat on the town by the volcano. So I would just like to show you a very down to earth hazard analysis, considering the possible impact of Vesuvio on Naples. First you have to think that Vesuvio is downwind with regard to Naples. So all the products will fall to the east, very likely. Then if you look at the most destructive eruptions, they are in the order of one cubic kilometer. Thanks a lot. And then you see that the volcano, since 1944, it has been quiescent. So no degassing, no seismicity, no surface deformation. On the other end, on the other side of Naples, to the west, we have this volcanic field. And this is Campi-Fermeri caldera. So it's a caldera. It doesn't have any prominent topographic fissure. In fact it's very difficult to distinguish that it's a large volcano. And indeed if you go here in the middle of the caldera, the people will ask you whether they are safe from Vesuvio. No, it's true. You have to consider also that the caldera is upwind with regard to Naples. So the products will most likely fall on Naples. And then the most destructive eruptions were larger than 100 cubic kilometers. And then finally on top of this the caldera has been restless in the last 70 years, 60 years. So this is just to say that we should pay a lot of attention when we deal with volcanoes. There are some volcanoes which are really not so evident as we may expect, but indeed they are very, very dangerous. So the message is that these are large, long-lived and restless magmatic systems which are systematically responsible for the most destructive eruptions on Earth. So I guess this poses an additional care when we study volcanoes that these systems are very delicate to handle, I would say. So caldera is defined as a subcircular depression with a diameter which is usually larger than that of a crater. And the most important part is that it involves some removal of magma from the reservoir. This may occur to an eruption or also in many cases through the lateral intrusion of magma at depth. And of course, which is the most distinctive feature of a caldera? A feature you will be able to recognize very easily is the topographic beam, which is a morphological feature because this lies well outside the structural rim where the ring fault is, and the ring fault is responsible for the subsidence of the central part of the caldera. So what we usually see is just the erosional part of this ring fault, and then the ring fault is usually valid. Hidden. And you may able to recognize it if you're lucky if there is some vent just above it because the magma is intruded in the ring fault. Then inside the caldera, many times you may have an uplifted area even of one kilometer over long time scales in the order of several thousands of years. And these are resurgence. So this resurgence exposes the intercaldera igrimbrite as well as any sedimentary deposit within. So this is the established classification of caldera which has been proposed. And this is based on million field evidence. And these are five types, five end members, I would say, of geometrical features of calderas. I don't know how much you know about this, but this is a very simple type. It's called a pistol because it's a coherent block which is subsiding. This is a piecemeal where you have different blocks which are collapsing. The trapdoor is an asymmetric structure with some flexure on one side and the ring fault. Well, actually this is not a ring fault. A fault because it's not all around the caldera. A fault on the other side. A down seg is a, well, less pronounced caldera. It is characterized by a broad flexure at the surface. And then we have a funnel which is a very narrow and deep caldera. So just look at these five types of caldera here and think which may be the most suitable to explain, for example, this caldera here which is not far from here, it's in central Italy, Bolzena, this is the map view and this is the section view. So which types of caldera would you think may fit with this structure? Well, okay. I think there is plenty actually trapdoor. Yeah, it's a possibility, but of course this structure looks like a piston. Okay, so the central part may resemble a piston. On this side we have a down seg, so we also have some down seg component. On the other side here, which is here and here, we have lots of faulting, so it may be a piece mill. But yeah, maybe the overall structure is a trapdoor. So if you want to make all these a piston like caldera with a down seg dream, a piece mill dream and an overall trapdoor structure, which is too complicated for me. So you see, in this case we are not really able to explain the structure of this caldera just using these five, the simple five geometric types. And this is not just because this caldera is difficult to be understood. There are many calderas which are like this. Actually, possibly most of the calderas share this complexity. And then there is another problem which is about wing faults, the nature of the wing faults. Cindy was talking about tabo in New Zealand. This is, well, a very simplified structure of tabo from grammometric data in map view and in section view. And you see that the caldera here is bordered by these inward dipping faults. These are thought to be normal faults because of the displacement. And we have a similar stuff here also in bias. If you see the subsurface structure, all these faults here are inward dipping normal faults. And of course nobody has ever been there, adept to see that those are normal faults. I guess in most of the cases we were able to think that these are normal faults because you have a drill or you have this layer which is here, in another drill or the layer is up here, so you just put some normal fault in between. And you know why? Because this, I think this is something which volcanologists have been relying on a lot in the past because the idea is that if you have magma, if you have a volcano, you must have magma. So you must accommodate the magma in the crust, so you need extension to accommodate the magma in the crust. So if you need extension, the normal fault is the best candidate to have this extension. But of course there is a room problem, a space problem here. Do you know it? For example, look here. If you want to collapse the central block, if this has to go down, what happens? You must extend to the sides. Otherwise this will not go down. And of course this depends on the deep of the faults, but if you have a collapse in the order of 1,000 meters, maybe this is as deep as 2,000 meters, then you need to have a few hundreds of meters of extension. And if the collapse occurs in one day, today as it's usual, then you need to extend the crust of hundreds of meters in days, which is completely unrealistic. So you have this important space problem with this inward dipping normal faults. Some people have been wiser, and so they were drawing the faults in a subvertical geometry, so this doesn't make any problem with the space problem, which is of course feasible. And other people have been more provocative. They were drawing the ring faults like this, outward dipping. And these outward dipping ring faults imply which type of kinematics. If this is the outward dipping and this goes down, it's an apparently reverse kinematics. And many volcanologists are quite scared about this reverse component, because they think that there is some compression, so this means that their magma cannot make it to the surface. So many people really do not accept so much this model. So at this point, I hope you are sufficiently confused to ask yourself the same questions, which are here. The first is how can we use this established caldera classification to interpret the structure and evolution of calderas. And the second is which is the nature of the ring faults bordering the caldera, and in particular how can we solve the room problem. So we go to the first part of the talk, which is concerned about the structure and the evolution of calderas. You know what is this picture? This is an earlier view of Miyakej Jima, caldera in Japan, which collapsed in 2000, and you can clearly recognize an innermost ring fault here, and an outermost ring fault. We will see this in more detail in a few slides. So the first thing I did to try to understand something about the structure and evolution of calderas is, of course, to go to the field. I tried to go to some nice, well-exposed calderas, not so much like Campi Fregre, where there is a lot of sea, there is a lot of post-caldera activity, a lot of people, a lot of buildings. So if you go to these very peaceful places, you may try to understand something about the structural calderas. But even actually in the best conditions, like this caldera here, where there is no vegetation, there is active, there are two lava lakes, well, actually now there is one lava lake here, but a few tens of years ago there was another lava lake here, so it's a very active caldera. Even in these very favorable conditions, you can just appreciate the shallow morphology, not even structure of the caldera, and you cannot say anything about the deep structure. And so you have a sense of frustration if you want to really understand how calderas work on the field. I try to use analog models. They are very simple and they are very, very simple. It's a good tool, and you can really understand the processes behind with these type of models. So I started to make these models. This was a long time ago, and the funny thing is that at the same time many other people were doing the same models, and we didn't know about each other. And then when we compared our models, we saw that all the results were very much consistent. And this was independently of any boundary conditions. We had different materials, different apparatuses, different boundary conditions. Everything was different, but the results were very much consistent. And I guess this is a very strong point, because if you have different boundary conditions in your model, and all the results are the same, it means that maybe there is something you should think about. Maybe the models are not so funny as they might seem. And so I'm going to summarize here the structure and the evolution of all these models. And this is just depending on the amount of subsidence. So at the beginning, if we have a very low amount of subsidence, we develop a down seg at the surface. This is stage one. And the down seg results from the upward propagation in the ring folds, which are still buried at depth. So when the ring folds reach the surface, we have a scarf. And this scarf is the result of an outward dipping reverse fold. So those people were right maybe they were right in pointing out this outward dipping reverse folds. And stage three then, if we further increase the subsidence, we develop an outer, which is related to the upward propagation of these peripheral folds. And when these reach the surface, sorry I have some problem with when these reach the surface, we develop an outer scarf so that we have two concentric ring folds. And the outer most here is an inward dipping normal fold. So while the inner most ring fold is actually we will see is the mechanical explanation for the formation of calderas, the outer most normal fold just results from the collapse of this wedge here, which is unsupported during the subsidence of this central block because this goes down so this has to collapse and it collapses with these normal folds. So now you may wonder which is the reason to have these reverse folds here. There is no compression actually, it's just a matter of a differential vertical movement. So if you have this type of movement this may be a caldera the flow of a caldera, which goes down then the sigma one trajectories are not vertical but they are archuate and so the result is that you need a fold at a low angle with these sigma one trajectories which must have this geometry and these kinematics. So you end up with a reverse fold which has nothing to do with compression. So I guess volcanologists can relax about this problem. The only minor difference I would say it's a minor difference for me is the results from the thickness of your over burden. If you have a very thick roof overlying your magma chamber your experimental magma chamber then you just have a repetition of the same features you see with a thinner roof. This is the first set of outward dipping reverse folds and then when they meet you develop another one and then you may develop another one and then outside you have these normal folds. But you see this is just a repetition of what you see here so it's nothing new actually. So of course this is all very nice and it may be interesting but which is the feedback from the field or from geophysical data are these structures realistic in maybe all the modelers have made the wrong move together. Actually if you go to the field the problem is that before you make the models you have a certain view then you make the models you start to have ideas and you start to look for those data on the field and actually you see that there is plenty of evidence for it and people have usually neglected it for decades. So this is for example this is the map view and this is the section view of a pit crater somewhere in Nicaragua and this section exposes the deeper part of this pit crater and you can see that there is plenty of outward dipping reverse folds and outside we have some subvertical or inward dipping normal fold which is very similar to the experiment. Of course this is a pit crater it's not a caldera but the collapse mechanism is absolutely the same. If you look at the the distribution in some calderas like for example Rabaur here is one edge or the caldera here is the other edge or Pinatubo like in 1991 this is the seismicity which was developed during the 1991 eruption you see that the seismicity is focusing along these outward dipping folds. Here the folds maybe are too deep they were just meeting at the surface so this may explain why the Pinatubo caldera actually is very very narrow. But probably one of the most interesting pieces of evidence for this for the presence of this type of structure also in nature is the Miakezima collapse in 2000. This is a geological map which is derived from the picture I showed you before you can see we have this inward this inner ring fold the outer ring fold and if we make a section that also studied this eruption you can see that this inner ring fold is an outward dipping reverse fold whereas the outer most ring fold is an inward dipping ring fold so this is very much consistent with what we see in the experiments and it's not just Miakezima but we have a very similar structures developed also during the other caldera collapses which have been occurring during basaltic volcanoes in the last decade like in Dolomje in 2007 or even in Iceland in 2014 so at this point we see that there is a consistency between what we see in the experiments and what is there in nature so at this point if this is true we may try to make a further step we can say okay let's try to see what we have at the surface which is the structure of a caldera at the surface and try to extrapolate using the models which may be the structure of the caldera adept so for example if we have a down seg of the surface we can see that there is a a fault which is adept which is propagating upward and this doesn't have reached yet the surface or even if we see two pairs of nested calderas sorry, a pair of nested calderas we can think that we have an outward dipping reverse fold and an inward dipping normal folds so just looking at the structure at the surface we can try to say something about the structure adept using the models so in doing this here we listed all the calderas all the best known calderas with their diameter their amount of subsidence and in particular we assign to each of these calderas here a stage which corresponds to one of the four stages which we see in the experiments so for example miagezima is a stage four caldera because it develops the two ring folds and if we plot all of this on a diagram this is the subsidence and this is the diameter and you can have different type of stages with different colors all of this may look chaotic at a first glance but actually if you help yourself with some lines which should be interpreted as broad transition zones not sharp boundaries you can see that usually below this line we have a stage one calderas down sex here we have a stage two calderas here we mostly have a stage three and here we mostly have a stage four so this means that despite of course some discrepancy because nothing is ideal in nature if you have a fixed diameter of a caldera and you just increase the subsidence so you pass from stage one to stage two to stage three to stage four calderas which is exactly the same as we observed in our experiments so this means that this apparent random distribution of the structural calderas as an explanation based on our models which is essentially for a given diameter a function of the amount of subsidence so there is a continuum in these evolutionary stages which is defined by the ratio between the diameter and the subsidence so the lower is this ratio the more evolved is the caldera and actually we may also try to go further because if we have this diagram here once we have a certain amount of subsidence for a caldera with a given diameter we just may try to infer its structure without even looking at the caldera that's a bit extreme of course but it seems to work with for example with the Bardarbunga caldera which was formed in 2014 and I guess that toaster will say something about this in the next hours Bardarbunga had collapsed in the other 65 meters the width of the caldera is 7 km and so this would place Bardarbunga here in the stage 1 calderas in the down-seg calderas and according to the models they also interpreted the formation of the caldera as a down-seg where the ring folds at depth they didn't make it to the surface because the deformation was the subsidence was too low to develop the folds up to the surface so at this point you may also try to understand that and choose these experiments as well as their comparison to nature for an updated classification of calderas which just not takes into account only the geometric features of the calderas but it also takes into account the structure and mostly the development the evolution of calderas and this is just using the amount of subsidence for a given diameter passing from stage 1 to stage 4 the difference between this classification is almost like looking at an image in one case and looking at the movie in the other case so we may go back to Bolsena and rather than saying well, Bolsena is what did I write here asymmetrically collapsed piston type caldera with down-seg and piece mill rims actually it's just an asymmetric stage 3 caldera because it fits somewhere here in the previous diagram and we also use these models to try to understand something about the kinematics of collapse and to do this you can use this particle image velocimetry technique which tracks the motions of your particles on the surface of the model and in section view for example this is very high aspect ratio caldera and what you do you just there is a glass and you can track the motion of all the particles which are subsiding along this vertical line here from the surface to the depth and you can track this motion from the surface to the depth as a function of time and so you end up with different amount of well different rates of collapse so you can have some velocities and you can distinguish different types of collapse like a continuous collapse here the velocity is pretty much constant and it is similar to the one which is the same actually to the one which we impose at the base of our sand pack because in this case this is sand but then you may develop at one point some sudden collapses where the velocity may increase up to four orders of magnitude just keeping our source at a constant subsiding velocity and then after this over in a background of constant, quite constant collapse these dark blue areas you have spikes of higher velocities which give you an incremental collapse and this occurs when you develop the ring folds and you have some friction along the ring folds and this incremental type of collapse is very much consistent with what we see in nature in the last the last collapses for example at Miagezima again or Dolomje using the tilt data like this is Miagezima and this is Pitonella for next or even the seismicity data so this is interesting because in this case you may even be able you have some more quantitative results and you may even be able to match your experimental results with geophysical data which support in both cases an incremental collapse of calderas usually do not collapse in a continuous way but they collapse step by step so now let's talk about the second problem which is unrest unrest at calderas is a very important and delicate topic and I don't know if you are aware about unrest unrest can occur at any volcano and this is a deviation from the baseline in the geophysical and geochemical indicators so it's usually highlighted by surface deformation gassing, gravity changes and seismicity this may be related to the activity of the magma chamber or even if you are below calderas you usually have another thermal system which may also be active and may also complicate your understanding of what's happening the important point is that the unrest can be eruptive or not this means that every eruption is always preceded by an unrest but not all the unrest episodes culminate in an eruption so unrest is unnecessary but not sufficient condition to have an eruption and the biggest challenge of volcanology is when you are in an rest is to understand whether this will lead to an eruption or not this is the challenge that volcanologists have to face in the next decades so this is a cartoon this is just simplifying things this is a possible way to develop an unrest to trigger an unrest but it's not the only way of course in this case you have the placement of some shallow magma this is deforming the hostrock so you have seismicity fracturing the fracture may increase the permeability of the hostrock so you may have that the the magmatic gasses go up and fill the thermal system may also induce air quakes and induce surface fracturing with a net result to have degassing surface deformation and also seismicity at the surface so there are different types of arrest I'm going to show you just some very basic type some reference type this is from Marco actually you can tell this much better than me actually this is a typical unrest episode of a basaltic caldera and what you have here the deformation on the top of our caldera is given by this red dot c you have inflation, then you have the eruption you have a sudden deflation and then you continue with inflation then you have eruption, sudden deflation and then inflation again this is something which has been already shown by Paul and Michael the second day and I guess it's a very typical pattern of many basaltic calderas there will be some complication which is related to the seismicity accompanying this deformation if the magmatic system is open you will not see any seismicity or you will see minor seismicity because you really don't need to fracture the ostrok the conduit is open if the conduit is closed you will have much more seismicity so this may be some difference in basaltic calderas but the situation is much more complex in felsic calderas so this is an example from Campi Fregre you remember Campi Fregre and you can see from the surface deformation that since 1950 the early 50s we have at least one, two, three and four uplift episodes four unrest episodes and this one in particular in the early 80s has been quite intense you had two meters of uplift in two years with a lot of seismicity strong seismicity and also some important degassing and now the current unrest is more subtle well at least the uplift is more subtle, gentler and we only have 30-35 cm within the last ten years we have minor seismicity but we have a lot of magmatic degassing so as you may understand civil defense is very much worried about the possibility of an eruption at Campi Fregre today because of these fissures here what do you think about this? any idea? you may wonder there was nothing here, no eruption when I had two meters of uplift a hell of seismicity so why should I worry now that we have just a few tens of centimeters of uplift and no seismicity so this may be something which tells you to relax not to worry too much of course this may be true but if you look at Rabaul what happened at Rabaul? exactly on the same years in the early 80s where we had the unrest at Campi Fregre we also had an unrest at Rabaul with approximately one meter of uplift and then a very strong seismicity then after this we adjust well stable ground level without so much deformation very minor seismicity and then after a decade in the early 90s we had some minor uplift 20, 30, 40 centimeters no more than that with some minor to moderate seismicity and then at this point the galdera erupted and it was a bi4 eruption so it was not really a minor eruption so this is quite I would say non-linear and you had a helier and nothing happened and you just had a minor perturbation of your system here and this was leading to an eruption the possible explanation behind this is that you don't have to think about unrest episodes as separate entities so you should look at the cumulative history of the galdera here so this unrest episode was probably priming the system while this one was triggering it was just a so here the magma chamber was really getting somehow the magma chamber the intrusion was getting to shallow levels and so you just needed a small trigger to induce the eruption so you may understand how important and how delicate is to study the unrest processes at galderas and there has been this important very important monography by Newell Zulithin about unrest at galderas it was a review about all the unrest processes which have been occurring before 1988 so we tried to update these unrest episodes the knowledge of unrest episodes after 1988 so we created the database with all the available data on unrest instrumental unrest episodes after 1988 and we were able to identify these 66 unrest episodes from 42 galderas which have been quite well monitored in the last decades of course this is just a small part well not as much it's a part of all the active galderas which are around in the world but many of them are not monitored so you really don't know whether there's been an unrest or not there in the last decades you don't have any information but the fact that in these 42 galderas we were always able to see at least once an unrest in the last 25 years suggests that also the other galderas may have experienced some form of unrest and this also suggests that the unrest may be the rule rather than the exception at galderas and in this case most of the unrest episodes were eruptive and most of them occurred at mafic galderas so we try to to identify some types of galdera unrest in our first qualitative approach just considering the composition of the magmatic system very roughly mafic or felsic and then the amount of opening of the magmatic system so plug galdera means a galdera which doesn't degas semi plug is a galdera which degases and open galdera is a galdera which degases and has an open conduit to the surface so you see that these mafic galderas they are mostly characterized by eruptive unrest as we saw before so this means that there is a frequent release of magmatic eruptions and this may occur with a typical pattern you saw before so inflation, eruption sudden deflation and again inflation the only obvious difference being the presence of seismic or not so that in plug galderas so you expect some seismic whereas in open galderas you don't expect seismic the situation is more complex as we have seen with felsic galderas we have a more complex behavior the release of magma through eruptions is much less common and in many cases we have this hydrothermal system which is complicating your understanding of the galdera sometimes it's buffering the signal the magmatic signal sometimes it's amplifying it we were able to define two main behaviors one is for plug galderas so non degassing galderas these are usually smaller actually we have very few examples of this type of galderas pinatubo may be the best example even though there was not an evident galdera on pinatubo before 1991 pinatubo is half way between a stratovolcano and a galdera I would say in this case we had isolated and rapid unrest episodes whereas in the other case in the other category which includes most of the galderas and these galderas are degassing galderas they are usually larger and they are continuously restless over decades or in some cases over centuries and a nice feature is that these galderas are usually associated with resurgence so this is the relationship between all these features should be better investigated also to understand resurgence of course all of this is very qualitative so we tried to do a quantitative analysis so a statistical analysis of this data and I guess this has been done with Laura Sunday and I guess this analysis is pretty much finished now so I'm just going to anticipate some result very briefly in a very short way I guess one of the most interesting results is that the eruptive arrest is shorter than the non eruptive or failed if you like unrest so this means that if a magma is ready to erupt during an arrest it will erupt soon and soon means within a very few months is this blue frequencies here if we wait too much it will probably not erupt and in particular in non open calderas so this means in calderas which are degassing or not degassing so they don't have an open conduit we see that this short eruptive arrest is always accompanied by high seismicity and degassing and you may think this is complicated maybe not useful but just think about the ongoing arrest episode the ongoing arrest episode has been lasting for years maybe a decade and it has just some minor seismicity so if we should use this data and we want to really say something so we can if nothing changes in the while we may expect that the caldera campiferae caldera should not erupt in the near future we don't have high seismicity and the eruptive arrest is much longer than the one which is shown here to erupt of course this is a very simple analysis because it's just describing behaviors but what we have to do is to understand the processes behind so I guess this is an important the first important step to update that to understand how things may work in nature and then based on this we can try to develop some model which is trying to explain these observations some physical model so once you have an arrest there is usually some magma ready possibly ready to make it to the surface so you may expect to have some magma transfer so the next chapter here we will talk about the transfer of magma within calderas and you have seen on the other day the lectures of Claude Jopard and he was talking about loading of volcanoes so loading the role of topography actually with calderas we have the same problem what the opposite sign we have the unloading because we are removing mass we are removing crust so we may expect to have some some similar process but with an opposite behavior I would say and I guess actually that this unloading may be very important in controlling the stress distribution within a volcano and ultimately to control the path of magma transfer so as an example of unloading I will show you two processes two cases in one case we will consider a larger unloading which is given by a mafic caldera like Fernandina and in the other case a smaller unloading from a felsic caldera campiflegre so Fernandina is particularly interesting because it's quite a unique volcano because it develops these circumferential fissures outside the caldera rim and then which are here these black ones and then outside we have these radial fissures and these fissures have been observed only at very few volcanoes and these are mostly in the Galapagos islands so the problem is to understand this complex eruptive pattern and we have to think that below which fissure there is a dike so this means that we are talking about dike propagation, shallow dike propagation here and there have been many models trying to explain this but we believe that the most comprehensive model which is trying to explain all the observed fissure is the one which takes a loading into account this is a study we have been doing in the Eleonora in Marco and here we report these are results from a numerical model this is the topography of the volcano along a Nazi axisymmetric section from the center of the caldera, outward and this is the topography again, this is the caldera area and here we are just considering the stress distribution which is related to the topography, to the topographic variations so these dark lines that's a sigma 1 and the gray lines here is the sigma 3 and you can see that below the caldera you have, just below you have a subvertical sigma 3 whereas if you move to the side this is better showing this schematic diagram if you move to the side you tend to have sub horizontal sigma 3 distribution so this suggests that below the caldera you are promoting the placement of seals because sigma 3 is vertical and as far as you go to the side you are promoting the placement of incline sheets and then of subvertical dykes which in three dimensions are circumferential dykes whereas if you are on the outer slope you have this circle here which means that you have an out of plane sigma 3 so you are promoting the placement of radial dykes and this is pretty much consistent with what has been observed at least during the last eruptions 2005 and 2009 these are the imaged intrusions during these two eruptions this is a seal and this is a dyke this is an incline sheet which is then twisting along a radial axis to become a radial dyke towards the other part of the edifice so this is just to say how important is to consider the unloading especially on the caldera pattern of circumferential fissures here and the load of the volcano to explain the pattern of radial fissures outside so the idea is that caldera loading promotes sealier and the proximal concentric dykes whereas the load of the volcano promotes these radial dykes outside so we may try to go further and put Fernandina in a wider context actually reporting here the length of all the circumferential fissures against that of the radial fissures and here this pressure ratio where there is the unloading pressure and the dyke over pressure so for example Fernandina has a high unloading pressure because you have a deep and wide caldera and you have a small dyke over pressure because in this case you have some dense magma then you have basaltic magma and the magma chamber is very shallow so you need relatively low pressure to inject the magma to the surface on the opposite side of the diagram we have for example top of our campifragre where the topographic expression of the caldera is lower actually it's not so evident especially in campifragre and you have a deeper chamber and a lighter felsic magma so you have a different behavior here in this case so this is just to say that it may not be only the unloading which is somehow driving the presence of this circumferential dykes outside the caldera rim but it may also be important to consider the depth to the magma chamber as well as the composition for example here if we have lighter magma, felsic magma I would expect this lighter magma to go straight up without really turning and twisting to the side of the caldera now let's look at this campifragre which lies on the opposite side of the diagram and to do this I will show you what may have happened just before the last eruption at campifragre in 1538 because the problem when you deal for example with a caldera which has not been erupting for a long time is that you really don't know how the magma may be transferred there is no evidence maybe if you have an intrusion you know that the magma is there but you don't know how it may reach the surface so if you are able to reconstruct the last eruptive event which is the one which is near most to you you may try to say something and well campifragre has a limited topographic expression because the topographic difference between the uppermost part and the bottom part which is here in the sea is not very few actually so it's not like Fernandina and also but there is still some possible unloading which is also related to the difference in density between the deposits the sedimentary deposits within the caldera and those outside the caldera so you also have to take into account for the density difference and so it may have some unloading but it's much limited than Fernandina and the interesting point is that all the last and rest episodes I show you before they are culminating in this area in Potswoli so the maximum uplift is here and these these rest episodes have been interpreted to result from the pressurization of a magma chamber which has this shape here in map view this extent this green ellipse here which is an oblate reservoir and this has been consistently responsible for all these unrest episodes and the nice thing is that if you look at the resurgence because we have resurgence in Campi Fegré the resurgence is testified by this last start Samarin Terras so this was below sea level and then it has been uplifted now and the culmination of this terrace again is in Potswoli so this suggests that the longer term uplift coincides with the shorter term uplift resurgence has been occurring in the last well at least this resurgence has been occurring in the last 5000 years so we can say that in the last 5000 years the system the magmatic system was probably there and it has been controlling the shorter as well as the longer term uplift episodes and this is the limit of the uplifted area the yellow ellipse the limit of the uplifted area of the resurgence and this is Montenovo so that's where the eruption occurred in 1538 and there are some very recent geophysical and geological data which suggests that Montenovo was fed by a daik, north south trending daik which is actually circumferential with regard to this to the present magmatic reservoir so what we did is to exploit the unique amount of information we have on Campi Flegre because we have several historical buildings and monuments along the coastline these black points here which have been quite well monitored in the last 2000 years I would say because there is a wealth of historical, archeological information as well as of course geological paleontological and geodetic information so that it's possible to reconstruct the apparent sea level variations in correspondence of each of these benchmarks in the last 2000 years we have 20 benchmark along the coastline and actually it's not the sea which has been moving because the sea has been pretty much stable in the last 2000 years so we have to think about the ground movement so the coast was going up and down this is an example from Serapeo do you know Serapeo? Serapeo in market which was submerged until the early 60s and then after the latest uplift episodes today is about sea level and you can reconstruct from this and other monuments the ups and downs of the caldera in the last 2000 years and this is summarizing these diagrams I will not go into the details but the message, the summary is that just well a few centuries before the eruption so from 1251 to 1536 there was a strong uplift which was culminating again in the Puzzoli area 14 meters of uplift culminating here so this was very similar to what we have seen on the short and on the longer term but in the two years just before the eruption the uplift was shifted here in the area of the future mountain of event so the maximum uplift moved here and of course in this analysis we tried to remove any surface deformation which was related to the days immediately before the eruption which may have been induced by the emplacement of the dike so we did our best to do this and if we try to invert this deformation using these simple elastic models we end up with two sources one is just south of Puzzoli and within one error within some error I would say the longer term source and also the shorter term source which has been inferred to lie below Puzzoli this is at 4 kilometers of depth and then we have another source an eccentric source which is just below Montenovo at a similar depth this is the fit between the model and the data for the central source this is the fit for the eccentric source and this is the conceptual model behind what we see is that the magma has been accumulating within the central magmatic system for centuries just below the center of the caldera and then at the last minute two years before it has been laterally transferred to an eccentric reservoir and from here a dike a circumferential dike was propagated so we have this lateral transfer of magma which we may interpret to result in the loading of the caldera from the moderate unloading of the caldera which is induzing this stress distribution the one you saw before from a vertical sigma tree below the caldera to an inclined sigma tree to the side which is induzing the lateral propagation of magma and then the vertical propagation of dikes and the interesting thing is that it's not only this eruption which may have followed this behavior if we plot the location of all the vents in the last 5,000 years from the center of the caldera radially outward we see that these vents they peak at 3 kilometers from the center of the caldera so this shows that in the last 5,000 years despite the maximum uplift has been systematically occurring in the center of the caldera magma was transferred laterally outside, immediately outside the most uplifted area this is the limit of the resurgence so just outside the resurgence and within the caldera folds so this suggests that we had this mechanism we had during the mountain overruption may have been occurring also in the last 5,000 years so it may have been a more general mechanism which is due to unloading which is able to migrate the magma laterally eventually erupting outside this most uplifted area and it's not just campifragre because if you start to look in the literature also in this database of rest episodes in the last 25 years there are other rest episodes which share a similar behavior for example Rabaul there is Azo there is Okmok and also Toya so this may be a more general mechanism which is applied to many calderas so what we have seen is that the role of unloading in calderas in two different cases in one case where we have a large unloading we have a magma transfer outside the caldera rim in the other case where we have a smaller unloading we have a magma transfer within the caldera but still outside the central part of the caldera and this is interesting because it's also important for the assessment because if next time if you see a large a very large uplift in the center of a caldera I don't have to expect necessarily to have the eruption there the eruption may occur in a peripheral way in a peripheral area of course this is a very schematic approach I would say because we may have a lot of complications for example regional tetonics we have some extension or compression this will affect our stress trajectories so the height path will be affected as well and as I told you also the composition depth to December should also be taken into account so the final message about these lectures the first is the importance of the amount of subsidence for a fixed diameter of a caldera in controlling the structure and the evolution of a caldera second is the importance to understand the unrest processes which seem to depend on the composition and the opening of the system and the challenge is to try to understand in a physical way the observations we have from a statistical point of view and finally we have seen how unloading may play a different role in different types of calderas that's it, thank you very much