 Okay, let's get going. So welcome everybody to this CSDMS webinar. Today I'm really thrilled to introduce Dr. Tristan Saul from the University of Sydney. Those of you who know Tristan's work will know that he is an absolute pioneer in coupled modeling of landscapes and seascapes, what's happening onshore and what's happening in the stratigraphic record. And was recently a pioneer of both modeling surface processes in biodiversity and global landscape evolution modeling. So I think he's gonna tell us something about some of those things at least. And so without further ado, I'll turn it over to Tristan for Landscape Dynamics Dictate the Evolution of Biodiversity on Earth. Hey, thanks Greg and thanks for the invitation. So let me try to share my screen. Let's go. I will get rid of all the zoom thing. Okay, so can you all hear me? Yes? I can't see you now because I get rid of all the things but I will say, I will take a no answer for yes and. So yeah, so thank you for having me. I didn't realize that it was going to be 1 a.m. in Sydney where I'm sitting now but that's what happened when you are naturally organized. So that's fine, I'm awake. So I'm going to talk as Greg was mentioning about some of the recent work we've been doing looking at global scale landscape evolution model and trying to make some links between what we learned from this simulation and the evolution of life. And so this work basically we've been developing the code for several years and we start to make or some outputs now that we can tell something about since maybe a bit more than a year now. So I guess I've got like some introductory slides here. Pretty much everyone is going to be aware of it but I start with it. I took this idea of the analogy with the human body from I think I read it in one post from John Perron and I really liked it. So I took it for myself and so basically the idea is that the Earth's surface is the living skin of our planet and this surface is basically at the interface between two main processes. So place tectonics and surface processes and by surface processes I'm specifically going to look at the effect of water and gravity and not really talking about wind and ice. But here you've got two examples of this effect of this plate and surface process, plate tectonic and surface processes with the Himalaya mountain on the left and the Lena Delta in Russia. And so really what I'm going to talk about today is basically what we can learn from this model specifically when we think about biodiversity and I could not have started I guess this presentation without this beautiful diagram from Vernon Bolt from the mid-19th centuries where already like this idea of relationship between landscape, geology and species distribution and diversity is well illustrated. And so this idea is coming back from several centuries but still we are not able to really integrate that in a kind of a modeling framework and we are working mostly qualitatively towards understanding this thing. So specifically what was the question I'm going to ask is why do we see areas where we've got on the earth surface where we've got a lot of biodiversities where and also how did basically as the earth surface environments have potentially changed the course of biodiversity and life on earth. In many of the approach that have been used in the past, most of the time people have been looking at landscape as a kind of a steady thing. So stable geomorphic features and especially when people have been looking at landscape in deep time they have considered that this landscape is pretty much static. In reality and most pretty much every geomorphologist knows that this is a dynamic system and that it can evolve really rapidly and change and it's transient by nature. So in this talk I'm going to basically have two main components. The first one is going to be talking about how we could improve on paleogeography reconstruction using some landscape evolution models. The idea is that what I'm going to present is still work in progress. There is a lot of things that could be done to make better prediction or better reconstruction but it gives you an idea of potentially what can be done. So that would be the first part and then after I've got a second part which is more related to what can we learn from these models in terms of the interaction between the Earth's surface and biodiversity. So let's start with the first part. So the reconstruction and trying to define and build a tool to basically do some quantitative approach to simulate as this paleogeographies. So when we think about Earth's past landscape usually we are dealing with maps like this one so the one on the left here which basically are created from like looking for example at analogs but looking at present day and trying to make some inference from that. Also based on plate reconstruction, plate tectonics reconstruction and also of course based on a lot of geological data so it could be fossil records, it could be stratigraphy but usually what we end up with is a map like this where you've got over time pretty much like kind of polygons with a mountain range for example here in orange you will have like a mountain range that forms and potentially disappear over time and this paleogeographies are relatively scarce and for people working with landscape evolution models there is not enough complexity in this landscape to really be useful for us. You've got on the right an example which is a bit more refined in terms of texture of the surface. This is some proprietary software where you can have this kind of data and you can see how basically they use different drilling sites different fossil records to try to infer the position of the elevation over time and on this specific one here they've got also some kind of representation of the main drainage system and so in both cases basically this map they are relatively coarse and in terms of resolution they don't really have a lot of information but anyway they are mostly used a lot by the community specifically to reconstruct atmospheric and authentic climates and also to understand like the distribution of natural resources on the surface of the earth but also used for evaluating the evolution of life so when people are dealing with meta-community models or models of representation of the competition between spaces at macro scale basically they will rely on these maps to basically build their models. So our idea was can we move from this kind of qualitative understanding or maybe not move from that but can we use some tool to see how well as this qualitative representation of the earth or how far they are from potentially the reality and so what we try to do is to design a tool to be able to test as this paleo geographies. So the idea was okay we have some understanding of how we should simulate basically the evolution of the geomorphology of the earth you know what we do, what we usually do as a landscape evolution modeler we will take a specific region we will get some idea about the precipitation, the tectonics of the region and then after we'll have a guess on the initial surface and run a model forward in time and look at the evolution of the erosion and deposition and our idea was to do exactly that but for like instead of working at catchment or regional scale to upscale that to global scale. So here you've got on the top here a paleoclimate models which depict the rainfall I guess you've got some tectonic and in our case we are dealing we are not we are accounting for dynamic topography and we can combine the two on initial paleo geography and run our landscape evolution model to get an idea of the amount of erosion and deposition and also on the geomorphic evolution of the system. So to do that basically we use a relatively standard model which basically will account for change in elevation as a function of uplift, erosion deposition induced by rivers and ill slow processes. So something which is really standard to all know about that. So for the erosion rate for the erosion rate we use stream power low and we've got vertical deposition rate which is dependent on sediment flux versus the water flux. And so really like the main advance I guess in what we've been doing is basically trying to develop ways of running these models on supercomputers to be able to get like relatively high resolution at global scale and when I say high we are talking I think the two simulation I'm going to show one is run at 10 kilometer resolution globally and the other one at five kilometer and to do that we had to work a bit on how we basically solve this equation here and usually when people are developing landscape evolution models they will pretty much design their tool around like a kind of a graph theory because like the river network they land themselves quite well in this kind of graph theory and what we did here was to use a matrix-based approach to be able to leverage on some existing tools that basically can be used over like a multiple processor and are quite fast to solve. So we rely heavily on PETC in our model. And so the name of the model is called GoSpell there is nothing religious here it's basically GoSpell and the approach that is used is really like a standard landscape evolution model so you will give as input a set of parameters so the elevation, climate, tectonic and you will get as a result of your model a series of outputs that you could use to investigate landscape dynamics but also river, water and sediment flux from rivers and also you've got the ability to store stratigraphic layers over time and so you can potentially reconstruct a base in evolution and of course like in any other models you will test your initial input versus empirical data and potentially have to basically look back and forth between these inputs and the empirical data based on the outputs you've generated. So GoSpell in itself is initially tailored for global simulation but can be used also at regional or even catchment scale if people are willing to do this at this scale. So in terms of paleogeography here you've got at the top like a different sketch of one of the highest resolution paleo elevation models so that's the one from Scrutis and White 2018 you've got here the north of South America the northern Atlantic with the U.S. and I guess a bit of Africa here and here this is the East African Rift and what we can do with the model is basically run for a set of specific precipitation and tectonic forcing, run the model and look at how this specific landscape is going to be dissected by surface processes so you can see here for example you've got the Andes that's what the paleogeography is looking and that's after you run your landscape evolution model and you can see that in different regions here and the idea is that what we believe is important specifically if you are looking at a specific evolution and how some spaces are going to adapt in different regions the fact that you can account for some complexity in the landscape for example this deep inside valley that you can see here that will potentially have a huge impact on this distribution of biodiversity. So to run this model actually if we look at the top and the top image and the bottom one what we do here we are actually running the model for a specific time interval so let's say we are at 150 million years we are going to basically run the model with a set of rainfall conditions and tectonic up to the point where we reach a dynamic equilibrium and once we reach this dynamic equilibrium basically we will consider that this landscape is representative of the specific time slice that we are looking at so in terms of a hypsometric curve if we compare the two landscapes there the top one and the bottom one they've got pretty much the same distribution what change is basically the fact that we are integrating in this bottom map here a bit more complexity in the landscape but overall they match themselves in terms of distribution of elevation range so this approach which consists in basically running the model for a given period of time until we reach a kind of dynamic equilibrium is an approach that is routinely used for paleoclimate models so usually what they will do they will have a specific landscape and they will have some inputs I don't know let's say CO2 concentration in the atmosphere, a specific volume of water and they will basically run their model until they reach a kind of steady state and they will consider that also as representative of the atmospheric and authentic climate for a given time so we basically use the same approach here but to simulate the landscape evolution so with this model what we can do is basically create like this I think this is 5 km resolution model so we can run at different time slice what we should expect in terms of change in the morphology of the landscape but also we've got information and that's what you see here superimpose on this landscape is basically the drainage network the large places which are like kind of till blue here they correspond to place where you've got some depression in your landscape so it might be lakes or it might be just like a really flat region we don't really incorporate at the moment the vapour transpiration laws in our model so I will tend not to think that all these things are lakes they might just be dry lands and really a flat lands so what we do is we can run these models like this over time and get an idea of the distribution of the flow network over time but also like as any landscape evolution model we can also look at and that's what you've got here on the right you've got the distribution of erosion deposition rates and so you can basically start to integrate these things over time and evaluate the evolution over several hundred of million years of your surface and you can try to quantitatively test basically how your surface is responding to this tectonic and climatic forcing and so here I put this here because I think this is a good thing to note to really pinpoint the fact that this is not like there is still a lot of work to do so the way we basically in all the models I'm going to present here the way we are calibrating our model is basically we use present day present day condition and look at present day denudation rate and use that to calibrate our coefficient of erodibility in the stream polar loop once we have calibrated this coefficient of erodibility which is in our case uniform over the entire surface of the globe once it's calibrated we basically don't change the erodibility coefficient and assume that this is the same value for like the duration of the simulation so here for example for the we look at the entire fanerozoic we basically calibrate our erodibility at this stage and then after we don't touch at present day and then after we don't touch it for the entire duration saying that it's not a technological problem if we wanted I mean and there is a capability in gospel to actually have different maps with different erodibility as well as the capability to change erodibility coefficient over time the problem is that we don't have a lot of information to constrain this erodibility parameter over like deep time but you know like in the future that's something that we would like to do a bit better but here so what you've got is in black here you've got the denudation rate which is predicted by the model and if we want to compare it with some data we've got like here some data from Wilkinson based on Wilkinson and McEnroe based on the stratigraphic record and for close to the new gene or present day we've got data which is available which there is a bit more data that is available which help us to calibrate a bit better the model you know on this first 20 million years but we see that there is periods where in terms of trend for example for the last 200 million years overall the stratigraphic record is predicting like an increase which is also find here we see that during that time here between the ordovician and the Triassic there is a huge difference between what you can expect from the stratigraphic record and what the model is predicting what we think is that as further back in time you go the less reliable the stratigraphic record is because you've got the surface processes have this tendency to erase basically previous history as you progress in time so here is basically one of our models over the last 100 million years where we use these paleo elevation maps we use the paleo climate and in this case we use this rainfall which has been run on these paleo maps and we basically run our model and this time instead of waiting for the dynamic equilibrium to be in reach we run this model continuously so what we do is we run this model over 5 million years we run our model against the next paleo map from Scotty's we look at the mismatch between the two and we basically loop back and forth during this 5 million years interval to make our model match as best as possible with the next paleo geographies so what it allows us to do is basically to have a kind of a continuous evolution of the landscape and we can after use this and start to look at how basically based on the paleo map that we use how well we are able to reconstruct some of the stratigraphic recall so to give you an idea of what we can do we can for example map as a different catchment we can look at the size of this catchment we can look also at the amount of water which is flowing to the outlet and here again I'm taking I guess this analogy with human body by saying that like the rivers in this case they will correspond to the circulatory system of the earth surface so they will basically drain our transport sediments from the source to the sink which is the ocean or like underic basins and what we can do is really try to map how this circulatory system is working and even though the model is global the idea is that you can interrogate your model for specific places and that's why we try to to get a resolution so in this case it was 10 km which is relatively high so we can basically say ok I want to know what my model is predicting for example for the Orange River in South Africa and so what we do is we will map basically the outlet based on our model the outlet of the Orange River over time and we will look at the amount of sediments which is flowing through this outlet and so here you've got the results for this simulation where you've got in orange the simulated sediment volume out of the Orange River and you've got in teal here the measured sediment volume so overall the model is working alright we've got like this big pulse in sediment which correspond to the first uplift of the South African Plateau around like 95 to 80 million years then after you've got a period where there is pretty much nothing happening and in our case there is a second or renew phase of sediment increase which is happening by 20ma to present based on the measured values it seems that it's happening a bit later and the intensity is not that high but the idea is that we can start to reuse this model to say ok if we are really convinced that the values that are provided are from the estimates basically this measure value there are estimates based on the amount of sediments which is deposited on the margin so if we are sure about these values we can start to test how we should change the paleogeography to make it fit better with the data that we've got so in terms of output I mentioned the catchments that we are able to evaluate of course we've got erosion and deposition and mentioned that as well and for this specific 100 million years of evolution we recorded as well the stratigraphy and so what you've got access to is basically a global stratigraphy global stratigraphy map which is you've got one stratigraphy layer for every million years so you should have about 100 layers you've got access to the stratigraphy so in the stratigraphy layer you've got access to the sediment thickness for any specific time interval of one million years and also the porosity which is just based on a simple law so if I go to some of this cross section here I took like the South America South of South America and so basically you can, if you want make a slice anywhere in your model and look at what model is predicting in terms of stratigraphy layers and also the structure of the stratigraphy so you've got an example here of section one here is on the Patagonian shelf where you've got like really thin elongated layers that are deposited you can see here we are more like around the northern part of the section one huge continental basin which is a Colorado basin you've got another example here where we cross from the Andes on the west to basically the deep deep South Atlantic margin and where you can see like for example the Salado basin and here's the Pelotas basin where you can potentially use some existing data as we looked into before with the Orange River you can for example look at seismic seismic lines and you've got a bunch in the region and look at how your model compare with the prediction in terms of thickness of the deposits and how well it's able to reproduce you know some of the different slice relatively interesting is like this section three here which cross north which basically go from the north to the south so you've got the north here and the south there and what we see from our model is that we've got like an initial part where this Colorado basin and Salado basins are actually connected and this is really induced by the initial paleo topography that we use and we had to wait for like half of the simulations about like 50 to 40 million years ago to start to have like kind of a late selection which correspond to a mountain range here which is called the Tandia High so if we compare what the model is predicting with some seismic interpretation we see that the model is actually predicting that this mountain range here is actually happening quite late in the paleo geography and we can say that because when we look at the stratigraphic record it really differs significantly from what we see from the seismic so our idea is that potentially we can use this type of simulation to change a bit the paleo geography and potentially have this Tandelia height that happened a bit earlier in the reconstruction so this shows basically an idea of how you could use this simulation to improve on the paleo geography that are available so to conclude on this specific part here what we are able to do is we are able to use this model to potentially give a bit more quantitative metrics on the evolution of the paleo geography and use this model to improve on existing paleo geography records so now I am going to talk about the second part which is how what we can learn from this model in terms of evolution of life so if I go back to the first simulation that I showed which was over the entire panheozoic what we did was looking at the position and amount of water and sediment that was flowing out of the continent so here what you have got Matt is basically the major river outputs which are flowing to the ocean so what we did was getting the total sediment flux that was going out in the ocean so that will correspond to this circle like purplish circle that you see and also map basically places where we have got sediment which accumulates on the continent so that will be the red areas on this map so if we look at how this evolves over time we have got in blue this is the total erosion flux what you have got in purple is the net sediment flux which is directly feeding the ocean and in orange is the amount of sediment which is actually going to be stored on continents for this particular time so what we have got is we have got an initial let's say relatively quiet period from the Cambrian to the Sicilian then after we have got an increase a kind of a plateau here and then after a drop just after the assembly of the Pangea and then after the breaker part of the Pangea we start to have this increase which basically will mimic some of the main mountain ranges that forms during the Triassic and we see that the continental positions they start to really kick off just after the Permian and through the Triassic and then after they remain relatively constant over the rest of the time so what we try to do then was trying to see if there is actually a relationship between the diversification of marine and terrestrial life and if this diversification is related to the physical environment we should be able to learn something out of this model so first thing we did was we are not biologists so we just took like the number of marine families over time and what we have got just after the Cambrian explosion we have got the gobes of the great Ordovician biodiversity events so after this steep increase we have got a plateau up to the big mass extension in the end of the Permian and then after you have got this steady increase from the Triassic basically to present day so if we basically on the top of this map we put our net sediment flux which is delivered to the ocean, what we find is that we've got like this plateau which corresponds to this seems to match with this initial plateau we've got a decrease which is before the mass extension from the Permian and then after we've got our increase which is shown here if we do a quick correlation between the two we find a really strong person correlation of 0.9 which suggests that there is a strong correlation between the two so one option is that this strong correlation is actually showing us the fact that as rivers are transporting sediment to the ocean they also transport nutrients which are the main, the primary food for organism and that's actually the fact that you've got this change in sediment flux to the ocean that will basically force the diversification of marine life so also we are aware that the fact that there is this correlation doesn't mean that it's actually a causality and we would like to think that this is actually a causality but at the moment we can't really tell that potentially the fact that we see this strong person coefficient is actually because the places where we have found some fossil records is actually places where there is a lot of sediments that have been deposited and that we are able to basically store these deposits so in this case what the model is telling us is that there is like a big bias in the fossil record and potentially what we can do is use this model to find some unexplored places where there is some sediment accumulations that have not really been looked into for fossil data another thing that we can do is now moving from the marine life to the continental life is trying to evaluate how these geomorphic features that we are able to map in the model can be used to evaluate plant diversification so what we did is we took some of the morphometrics from our models so you've got like the roughness of the landscape at the top, slopes, water flexes and we categorize them and combine them into one single matrix that we call the physiographic index which basically represents the complexity of the landscape in terms of all these parameters and what we did is we again looked at the number of spaces over time in terms of plants and tried to see if there was any relationship between what the model was telling us and the evolution of these plants so the first thing we did was mapping the total area of this andorraic basin over time and in this case what is quite nice with the plants is that you've got pretty much a linear increase and in our case total area cover of sediments also is increasing linearly so we've got here again a strong correlation between both curves and so on top of the total area which is covered by andorraic basins we also added like the component which is related to this physiographic diversity index that I talked about before so we looked at the variability of this physiographic index over time and we combined the total area covered with this physiographic variability and we end up with another even stronger correlation with the R square of nearly 0.9 which suggests that the expansion of plants only started once we started to have some soil some good soil condition for the plants to start to grow and basically it also facilitates the development of rooted plants and so the idea is that here again it seems that there is a strong link between the evolution of the landscape the transport of sediments in different areas and the evolution not only of the marine life so to conclude I guess the idea is that this is still working in progress our idea is to be able to add the surface processes component into like this global paleo earth system modeling framework so there is people who have been working on mental convection at global scale there is this paleoclimate models and the biodiversity which at the moment is using these two components of plate reconstruction and paleoclimate and potentially our goal will be to add these surface processes component on top of that to see how it might help to evaluate biodiversity over deep time thank you let me stop this can't see anyone thank you so much Tristan for such an incredibly stimulating talk so we have some time for questions you can either post a question in the chat if you like or raise your hand and I'll call on people an easy way to do this you raise your physical hand but I might not see it so an easy way to do it is if you click on the reactions button and the bottom of the screen you'll see there's a little raise hand feature is what it looks like when I do it so I'll start out with a question since I see no hands up yet one of the things I'm struck by in this and a question that came up is what about the role of weathering it's the weathering both chemical and physical of the rocks so this is the sediment chemical weathering that's created in drawdown of CO2 it seems to be highly sensitive to temperature and precipitation there's this idea that when you get like our continent collisions in the tropics especially if there's a lot of mafic material uplifted you have a strong CO2 drawdown effect and weathering also will provide the nutrients that drive the biotic system so have you thought about how you could incorporate weathering processes which we I think it's fair to say we don't understand really well the quantitative level but how could you think about that role in these global simulations yeah so there is one way of to be able to start to look at that is to have a kind of representation of the surface geology to some extent to know what kind of rocks you are exhumating so for example there is this work from Moosdorf where you've got like Imad basically the surface geology and he's got a kind of coefficient of aerodibility which is depending of the surface geology the problem is how you integrate this surface geology over time and again I mean in terms of like basic weathering it's not really a problem of modeling capability it's more a problem of how do you have the information for doing that and that's where the problem is in terms of where and I didn't mention that but one aspect that is I think important specifically for people who are doing paleo-climate models and because of the role that weathering is playing on CO2 there is like I think a lot of things that could be learned from this landscape evolution model and that could be applied to the paleo-climate community and at the moment most of the way this is dealt with in paleo-climate model is based on box model where you will have the distance to the mountain range the elevation of the mountain and from that you have the coefficient that is telling you that the amount of sediment that is eroded and so that's how they basically derived that kind of information from the landscape I think there is ways with this type of approach to improve on this and to go a bit further than the idea of using a box model and it's specifically it's really important I think also when you look at the places where these sediments are distributed because like when you just have like even like a meter change in your landscape you can have a river that goes from 2000 km difference in terms of where the outlet ends up so that's like there is a lot of things that are happening there Thanks Albert Kettner Yeah, thank you Tristan there was very inspiring talk it's just I'm thinking about having some sort of digital twins where you can turn off processes and see how different landscapes would evolve and the capability with your model is just beyond what you can imagine I think so that's pretty cool I got a question about your improving the resolution of certain landscapes you let your model you pick a time slice if I understand it well then you let your model run until it hits kind of equilibrium and there you know not much of the landscape will change at this point anymore but how do you know this is kind of a good representation of this higher resolution so that's one and then associate the question with it how much can you get higher in resolution before you're running into kind of you know maybe not realistic processes anymore where it's more maybe a statistical difference than really capturing processes do you reflect a little bit on those two? Yeah, so I will start maybe with the second one I think the resolution that we use here like one kilometer or five kilometer we use this because usually this kind of resolution they are considered as standard or even course resolution in landscape evolution model and we use the same kind of laws and the ones that are relatively course compared to some models but the stream power law in itself should work well in this kind of resolution after if we wanted we could have it's possible in the model to have higher resolution in some places that you are really focusing on and like Corsair One for example in the ocean if you don't really simulate landscape you could speed up your model by doing so to come back to the problem of statistically representative I think the idea is that here I'm just showing one or two simulations in reality like if we are using this kind of model at catchment scale we will run potentially 20 hundreds of simulations here we are even if the model can be run over on a supercomputer we are limited in terms of model time I mean if we are just talking about the model where we do the dynamic equilibrium that's fine because we are running with one time slice and a time slice like this at 5 km resolution it will take 2 hours, 3 hours to run on hundreds of CPUs so you can run multiple of them it's not really an issue and it's not continuous in time in this case that's why I guess also the paleoclimate people have been using this approach as well so you map something which is representative of your specific time slice if you start to run like the first model where I run it for 100 million years where it's actually a continuous model this took much longer to run than like running the entire fanerozoic with time slice every 5 million years because at each time step you have to stop your model we have an automatic process to do that but compare with the next time step from the Scottish paleo elevation check the mismatch, find what needs to be changed in your model and in our case we assume that what needs to be changed is the tectonic we don't rerun a paleoclimate model we just iterate up to the point where we reach something that is fine but in reality there is a need to really do a lot more simulation to be relatively confident with what we are suggesting but in any case even if there is still work to do at first order this kind of strong relationship that we find is quite... we found it quite amazing so that was the goal really of what we did yeah, no cool thank you other questions for Theresa? I'm going to toss out one more I guess seeing no others one of the things that struck me was that in your example from South America you pointed to an anomaly that the model was distinctly different from the evidence were there other notable... that seems like a good use of this kind of a model to identify anomalies where there is something that doesn't quite fit which may point to a gap in our knowledge were there other notable anomalies that you discovered in this process? many, but this tandelier height this is like a place of high endemism and so knowing this kind of things is really important but to give you another example like the Mississippi for example this is one of the best studied place on earth but still it's crazy like the amount of difference that we've got in terms of where where the catchment like the upstream part of the Mississippi DNH it was not working at all and after like in terms of where the main delta were depositing that was like really difficult I think every places where you've got really something which is relatively flat it takes pretty much nothing to go from what I was saying before from an athlete that can be 2,000 km apart like the Amazon was working really well that was good the Orange River as well but the Mississippi was an example where it was not working the Niger delta as well it was a bit problematic and this is because there is like I mean paleogeography is computed there is really a narrow kind of valleys at present day where this Niger river is flowing through which is not basically possible to simulate I mean to get at a course at 5 km resolution so you end up with something that is feeding your internal Africa basically so that's all these tricks where you could basically fine-tune the landscape fine-tune the paleogeography to get something better yeah awesome well we've hit the top of the hour thank you so much Jason for bringing up till 2am now in Sydney to share your work I'm awake now recording will be up in a few days if you want to catch it again or tell your friends thanks again