 Okay, everyone, let's begin It's my pleasure to introduce David Piles who's the director and technical research project Manager for the Chevron Center for Research Excellence at the Colorado School of Mines Happens to be a former student of mine So I'm particularly pleased that he's willing to give this talk and I will turn it over to Dave Well James for thanks very much for the introduction and I also want to thank Irina for inviting me to give this talk It is a pleasure to speak with you today about my perspectives on modeling and particularly as I am a field geologist and not a modeler But one thing I would like to do is walk you through some ideas that I have in terms of how measurements that field geologists make can can at least collaborate with and contribute to forward modeling I'd also like to thank my co-authors particularly Kyle Straub and who has contributed to a Part of this talk and then Doug Edmonds who were just starting off some collaborations as well I'd also like to thank some inspiration From some modelers such as Bill Ross James Witzke Irene Overeem and Michael Perch not to single anyone out But I think a lot of the work they're doing is certainly inspired me as a field geologist So what I'd like to share with you today is a review of some techniques that can be used to evaluate model and Outcrop uncertainty and test natural variability and sediment transport systems And I'd also like to share with you a vision of how field geologists and modelers can leverage from each other's perspectives And I show some neat images of some field areas for you up in the top is the book cliffs a great outcrop in Utah down in the bottom left is the Wasatch formation in Utah and then this is a great example of a of a deep water system that we're working in California so in terms of putting some context for you we all know very well that there's a number of ways to address stratigraphic problems and You know my world is here as the is an outcrop stratigraph for but Most of you in the room. I think our numerical modelers or you can therefore using numerical Simulations there's physical experimentation. That's an important way to work In addressing stratigraphic problems. There's obviously subservice studies I think the top images are pretty self-explanatory this bottom left image is a is an amplitude map a seismic amplitude map of Mini basin in Gulf of Mexico and then there's obviously land and seascape studies and what I really mean there is like modern systems and at least in my perspective and in in my experience what I see is very strong relationships between Numerical simulations and the land seascape or modern studies And I think though the previous keynote was a great example how the modern landscape is being used to test the efficacy of models and vice-versa And certainly there's a lot of research done relating physical experimentation to numerical simulations Now what I would like to share with you is some ideas that I have in terms of how Outcrop stratigraphers and numerical models can work together to address common questions but also to Understand to test the efficacy of models and you know the whole idea of what I'm basically saying is very Aristotelian Meaning that the whole is different than the sum of the parts in that if we gain knowledge from physical models Outcrop subservice studies the whole shebang We're going to be better off than if than any of the individual studies on their own Or if we had a number of small studies focused on a similar concept I could also cite James Savitsky, which is the more that we know If a numerical output matches that of a physical or System, then we know that we're going in the right direction And so that's what I want to share with you today is at least some of the perspectives I've had in terms of what how we could compare Quantitatively compare measurements from outcrops to that of simulated studies and again, this is just a perspective So there's a lot of work to do But to give a little bit of background on where I come from There's a lot of recent technological advances in data collection techniques in In-field stratigraphy that have yielded opportunities to better quantify Stratigraphic stacking patterns and spatial distributions of deposits of ancient sediment transport systems One notable example would be LIDAR. LIDAR is a detailed XYZ grid of Topography basically and what we've used in my team is a LIDAR machine We basically point it sideways at an outcrop so we can have a very strong or detailed image Of the surface topography the outcrop which allows us to then quantify Stratigraphic surfaces so this is an example from some work. I've done with a colleague Lorna Strachan and David Jeannette In Ireland just very recently. This is impressive in geosphere But the idea here is that you'll we're looking at a photograph on the top left of of an outcrop within this big study area that we did and then the LIDAR data image is on the bottom left and then on the right there is the surface topography of This system or of that local part of the outcrop. Well, this outcrop is highly complex. It's very large It's a few kilometers in in breadth and what we can actually do in a computer model is Track these stratigraphic surfaces and have detailed digits for how say a channel fill looks We can also register with a grain size distributions if we want to the types of physical sedimentary structures Which are a proxy for this the depositional processes and so forth now in this talk I'm going to be focused more on on surface stratigraphy just to keep it very to keep this as simple as possible Another other advance is our photogrammetry. This is an image of the book cliffs This is a this cliff is about 500 meters tall and what the what the way that these images are these models are made is by Crossing the focal length the focal point of a camera so that you can build like a complex three-dimensional image Literally out of photographs and again what this does it creates a complex three-dimensional image We can bring into a model and create surfaces for and that's again what I'm going to focus on in this talk is is surfaces And how we can use them so one way? I'd like to illustrate an example of how field geologists and numerical models can work together It's through the concept of Compensational stacking and that's going to be the majority of what I'm going to share with you today I'm going to show a vision of how we can do some stuff with deltas later So in terms of compensational stacking, let me speak as a stratographer here is Compensational stacking is a tendency for sediment transport systems to fill topographic lows And that's the definition that Kyle Straub put out in this 2009 JSR article It's an idea that we've been talking about for years though all the way back to Muti back in the 60s The this process of compensational stacking results in spatial changes in the locus of deposition by evolution So what I show here on this image is this is a generalized diagram of a shelf to basin profile Imagine it offshore Atlantic margin or Gulf of Mexico and the idea is that you have a shelf at the shelf And in this case, it's basically a low-stand or in other words at low sea level Transferring off into the slope and into the basin and what I want to focus on in particular on the show Basin profile or these distributive submarine fans which are common on I think all of us solicit clastic continental margins They can happen in interslope mini basins such as salt withdrawal basins in the Gulf of Mexico They can happen in the extensional basins such as those offshore, California And they can happen out on the basin floor and they contain channels and lobes that Form of radially dispersive pattern shown here is the seismic image the same one I showed earlier and this is from Rick Boboff study of the Brazos trinity system I'm going to use this to illustrate this process of Compensational stacking for you and what you're seeing here on this this seismic data image is basically it's it's Amplitude and the orangish colors of the hot colors reflect sandstone whereas the gray or colors reflect mudstone This is a map view and you see the scale bar on there Now what you see in terms of just a general pattern is that the channels are entering the basin on the on the right And they are forming a radially dispersive pattern where they transfer to lobes But critically this isn't just one time slice We're looking at the resultant stratigraphy that reflects an evolving landscape and in particular if you look closely You'll see I'm going to throw in these numbers reflecting lobe one two three four and five and the process or the Sequentiality of this is determined through superposition or cross-cutting relationships Which was derived from the seismic data But my point here is that that radially dispersive pattern that you see is a result of a number of time steps Let's look at this in cross-section to get a better idea of what this looks like and now I'm going to show you some seismic data offshore Corsica in Which you're looking at a shelf to basin profile where the shelf is up here Whoops the slope goes down and then into the basin floor And I'm going to show you this seismic line is just of one distributive fan in here And this is from Deptux work in in sedimentology and the idea here is this a seismic data image of lobes Which are those big paddle state things that we looked at and cross it in map view on the previous slide This is highly vertically exaggerated I think it's probably about a thousand to one vertical exaggeration just to give you an idea here but in terms of the sequentiality you can see lobe one two three four five six and you guys see how this Compensational stacking or the set of the tendency of the sedent transport system to fill topographic lobes is shifting the depth Of center of these units around through time So this is really exciting to me because what it's suggesting is that there's a there's a landscape evolution that can be derived from a cross-sectional field So in terms of where else does compensational stacking happen? Well, it happens in Delta shown here as an offshoot of the Mississippi River Delta in which Compensational stacking is certainly part of that system I think we can all imagine Delta's a lot of people have Delta experience here other systems are alluvial fans or What Gary Weisman we call distributive fluvial systems and this is a Landsat I mentioned what you see an older part of the fan Which is no longer active is shown in the darker colors Whereas the cool colors are the modern Modern active part of the active fan here and again the idea is that it shifted from the right to the left through time and Building this broad radially dispersive pattern Now this is cool. So but how do we quantify it? And this is where my co-author Kyle comes in and that Kyle in 2009 developed a metric for quantifying Compensational stacking and let me put this in context of where Kyle comes from well first of all He lives in Tulane But in terms of his frame of reference is what I'd like to share with you Kyle's a physical model and what I show here is a snapshot of his Delta basin on the right in which you can see the Active channel at least at the time that this photograph is taken is shown here on the on the right side of the basin Whereas the left part of the basin was abandoned Well during this experiment what they were doing is rising base level or in other words Subsiding the basin and this cross-section shown here on the left is what the stratigraphy the resultant stratigraphy looks like So now what's interesting when you when you're modeling things is that you have full control over Sediment flux Sediment accumulation rate or you can measure sediment accumulation rates You can measure substance rates is all programmed into the machinery and so it's no surprise that when Kyle Quantified Compensational stacking he was using this metric here, which is he calls Sigma SS in which the variables that's quantified are local sedimentation rate Long-term subsidence rate and that's scaled over the length of the cross-section He then uses that to develop an exponential It's a empirical fit of an exponential decay function now again I really want to focus on these variables It's local sedimentation rate and long-term subsidence rate Let me show you how this works just to develop an intuition that we're gonna use for the rest of The talk the idea here is if something's perfectly Compensational such as that on the bottom right image you note that the this channel levy system is Flying all over the place like a brick wall sort of beautifully compensationally stacked that would have a compensation index of one However, if the sediment transport system is sticky or in just a grading it would have a compensation index of zero and then this diagram here shows something in between the two So these lines on the graph are are going to become more obvious when I start to fit this with real data from an outcrop So now let's go back to the oh and by the way one thing that when I first read this article I was really floored by I love the idea of quantifying compensational stacking and one thing that I learned a lot in in this process and I want to give you some Insights in terms of how modeling and field geology work together is that there's a lot of insight that can be gained from modeling I never would have thought of quantifying compensational stacking had I not read this article and so Once I read it though it taught me that I need to start observing different things in the stratigraphy Which is why site Darwin's correspondence with a geological survey Which was every observation must be made for against some view if it is to be of service But if you don't have that perspective, how can you know to measure it? So now what I want to share with you is well I got excited about compensational stacking you'll recall the variables are are Sedimentation rate and subsidence rate, but now enter the field air the field geologies a field geologist world Which is something like what you see here, and this is a c-clip from Ireland So we know these rocks are 318 million years old We can roughly derive the sediment that these subsidence rate But it would be averaged over about a 4 million year duration, which is largely dissatisfying So when I called Kyle up I said hey is there any way that we can modify this and so we work together We spent a weekend on doing workshop, and then I went back out in the field and reevaluated some data But what we tried here is refining the input variables to measure Compensational stacking and so what we did is we refined it so that the numerator now is local deposit thickness Scaled over the mean deposit thickness between any two stratigraphic surfaces So what we're doing now is we're evaluating this this coefficient of variability for all pairwise combinations of surfaces that we could measure in an outcrop We have an exponential decay function That's going to fit a best curve and that will define our compensation index, which is the exponent so with that Theory I'd like to now describe for you how we compared the original method to the modified method again The whole reason for modifying what modifying it was so that it would accommodate for the variables that we can measure in the field and What I show here is again the same Tank experiment that Kyle put together in which the soup that the Graph shown here at the top uses the old metric that uses Subsidence rates and sedimentation rates and the bottom meant The bottom best fit curve here is for our new coefficient of variability Which is using local over mean deposit thickness and the bottom line here is that you'll see that the curve This the system starts at a compensation index of one and it transfers at smaller scales to about 0.57 using the first index and with our modified compensation index We have basically the same values although more error and I should point out that this these larger error bars that you see here are Due to the deposit measure the precision of the measuring instrument that we had we're actually entering the one millimeter territory They're on the left, which is the the resolution of the laser scanner that that was used to measure This so the bottom line I'm trying to say here is that we modified a metric from One that required omniscience of a of a system to one that requires just simple Ocular inspection or actually empiricism from an outcrop so we could have a common way of measuring an output of a model To that of a of a natural sediment transport system so now what I'd like to be do is Share with you that because this is a surface approach This is something that could be applied to any forward stratigraphic model I show here an output from Michael Perch's paper in 2006 much Michael Perch has been focused on event-based or rule-based models He gave a keynote here in two years ago. I believe it was is that right and So but the important thing about this type of modeling program is that he can track stratigraphic surfaces in the same way that 3d said flux I believe does as well and this is an image from Irene's article in 2005 and then lastly I believe this would also work for the stuff that Dalgadmins and his colleagues are doing as well And so the point is is if we can use surfaces and the thicknesses between surfaces We might be able to measure model outputs with that in actual systems So let me take you back to the outcrop and first what I'm going to do is use this methodology to address a question of how does Compensational stacking very spatially in a submarine fan or in other words if I compare the up-dip and down-dip of the system Are they the same and the second thing I'd like to do is describe how Compensation varies by scale or in other words if I were to look back into this cross section You'll note that lobes 1 through 11 are highly Compensational, but if I surgically extract and look deeply into Lobe 1 do you note that the seismic reflections are all superimposed on one another Indicating possibly that compensational stacking is a scale dependent process or in other words. It's now fractal So we can test these ideas using outcrops That's what I'd like to do now is take you to Ireland and this is the Ross sandstone system Which is about three hundred eighty million three hundred eighteen million years old it's nemerian in age if you're European it would be Pennsylvanian if you're American and The formation was deposited within Ponded distributive submarine fan and we've done a lot of work there in the past several years What I'd like to do now is zoom in to a closer view of this geologic map the area That's broadly yellow which is where the Ross sandstone crops out and shown here is a geologic map of the peninsula which is The yellow parts are the Ross sandstone Here's a couple images of the sea cliff exposures the thing I want to emphasize here is remember We developed a surface-based approach for quantifying compensational stacking so in order for this to be logically correct We better Be able to define Surfaces very well and what you can see on the sea cliff exposures that the surfaces are remarkably well exposed Here's another example of a sea cliff exposure there So what I want to do now is just take you to two different areas of this outcrop system an area called Rineville a point which is predominantly channels and an area called Kilbaha Bay Which is predominantly lobes so we recall on that distributive fan system that I showed you from the seismic image the channels Longitudinally transfer to lobes along their longitudinal transect and so that's that's what we're looking at We're gonna be looking at here. The next slide is gonna be just of the Rineville a point area So shown here is a cross-section that it took me about five weeks of field work to do this We started off with the lidar data image for what it's worth all the vertical lines on this cross-section are Where we have stratigraphic columns and the resolution doesn't show up very well But we're measuring centimeter scale stratigraphy here in terms of the cross-section. It's vertically exaggerated So at about I think five to one so we have 25 meters thick and it's about 450 meters wide The chat sediment transport system is directly entering into the screen So you're looking at a perfect cross-sectional view of the stratigraphy the yellow colors represent coarse-grain Sandstones whereas the earth tones are finer grain rock. Let me show you what this outcrop looks like So if we're looking here at the left side of the system You'll note here that you can see the left side of this beautiful This is channel 3 or C3 you can see all the bedding and stratigraphy in there If we look at the other side of the system, you can see a beautiful little pothole, right? Whoa, there's a mouse. There's a pothole right here And that's filled a shill class conglomerate and then there's a lot of structural sand above it But again, I want to emphasize how visible the bedding boundaries are Another thing I want to emphasize here is that we've color-coded the surfaces by the amount of erosion on them So you'll know on the stratigraphic cross-section. We have bold red lines those indicate surfaces that have a high degree of erosion and the the the blue lines are just have less and So what you'll note here is the system is moving around Quite a bit in terms of the the different channels are compensating a lot However, if we look within the channel what we see is significantly less offsets between the units and that's those are depicted with the black circles and Then at even smaller scales we see more vertical aggregation Let me we're going to compare that with this other system here Which is Kilbaah Bay and what you'll note here is that the system is much more tabular The stratigraphy is far more tabular and again the bedding surfaces are very well exposed And so for clarity what we see here is a lot of compensational stacking the system is shifting around a lot And what we're going to do now is take that and quantify it But just for clarification. We're looking at architecturally distinctive parts of the fan we're looking at up-dip and down-dip areas reflected by Kilbaah Bay and Pardon me by Ryanville appointing Kilbaah Bay respectively. So here's the model outputs basically What we learned here is that if we look at the graphs representing Ryanville appoint what we see is a compensation index Increases from small to large scale from point four three to point eight one in the same way That for the small to large surfaces in Ryanville appoint or pardon me at Kilbaah Bay, which is predominantly lobes We see the same process operating in that compensation is increasing with with size in The same token we see that compensation increases in the down fan direction So the point here is what we've done is we've measured the absolute compensation index for a stratigraphic system Which we can later use to compare with outputs from models We have not yet done that But what I'd love to do is work with Michael at some point in the very near future is to ask him What is the compensation index based on this rules-based model that you have and how does that compare with the natural system that we Have no I don't mean to infer that this system that I worked in Ireland Represents the natural variability of systems I have two PhD students are working on other submarine fans with varying sizes and shapes to see if the boundary conditions of the base And affect compensational fact stacking as well So the point is that we just have a methodology that can be used to address a number of questions We have an out we have a methodology also that can be used to compare Stratigraphic stacking patterns and outcrop with those from from numerical outputs at least that reference surfaces So another thing I'd like to do is propose just another simple idea in the last few minutes that I have and that is it dog admins Who's in the room here presented a geometrical model that evaluates how water depth and gradient Influence the stratigraphic architectures of Delta's so imagine a cross-section through this Delta going from up dip to down dip and Shown here is an output in or one of the figures from Doug's Doug's model here in which the Cross-section of the Delta is shown to spatially change from being a lot of top-set stratigraphy, which is shown here in the dashed Areas transferring to a lot of four-set stratigraphy. This is very intuitive But it's and I find it ironic that field geologists and stratigraphers for many years We're using this bottom model which basically shows no longitudinal transfer or change in stratigraphic architecture So the idea here is if you were to quantify the area above this black dash line here You would see that this is all top predominantly top-set stratigraphy above it Whereas that base and word or below it is but predominantly four-set stratigraphy or the mouth bars themselves So when I read this paper it gave me a lot of ideas in terms of how we can quantify stacking patterns in Outcrops and particularly one thing that I wanted to learn is how Shallow water deltas are those prograding into relatively shallow water have different architectural styles than their deeper water counterparts I'm going to show you an outcrop example of that in just a minute But the idea here is this is a diagram based on the geometry geometrical input characteristics that That's great the geometrical input characteristics that Doug proposed and the idea here is that the shallow water Delta Has a long component of its longitudinal transect that's predominantly top-set stratigraphy Whereas the deeper water counterpart has a comparatively lesser amount of top-set stratigraphy Therefore the stratigraphic architectural styles are very different and so right now we're working in the book cliffs This is a photograph from helper Utah And this is Mark Kershbaum who works with me at school of minds across section the heap put together And what we're working on is we're comparing this is the manka shale And these are a bunch of plastic wedges that are prograding into the into the basin in here This is about a 200 kilometer long transect through the stratigraphy the gray areas are marine shale yellows are coastline strata And all this other buff colored is is shallow or pardon me fluvial strata And what we're actually getting what we're doing in the field is we're comparing an up-dip Shallow marine system to its deeper water down dip component out here, which looks something like this So in the up-dip area we see a Delta that is prograding into about five meters of stratigraphy of water Whereas the deeper water counterpart is prograding into about 20 meters of water of water And what we're doing in the field is actually measuring the different facies types that we have there as well as the different architectural styles to first of all test admins as model, but then also be able to describe how heterogeneity how Different processes are operating in these two deltas of comparatively or reasonably similar ages And so then the idea here is that this approach can be transferred to and for models Not just this types that Michael Perch shows or the or the more process-based models that Irene and Doug have used But what we can do is test efficacy and uncertainty through these numerical methods that we're using in outcrop as long as they can be compared and transferred to model outputs and So in conclusion what I'm proposing here is that the more that we work together the more that we talk together That will enable us to develop common workflows as a means to relate Results of numerical modeling to natural systems and particularly outcrops in this example I think what this relies on is a conversation. So let's keep talking about this But another thing that besides model validation in terms of ocular It's it's an improvement over ocular comparison We have a quantitative way of comparing natural and numerical models But the other thing I'd like to really share is that coupled perspectives provide insight that can be used to develop new questions And I would say in both of the examples that I showed the Compensational stacking as well as the the Delta example These are perspectives based on modeling that I never would have had as a field geologist No, maybe other field geologists that are smarter than we could have done that But I needed the help of these numerical model outputs And then I can go out in the field and test them and then build upon them Hopefully and we can work together as a team to develop common workflows to address efficacy and and ask new questions So with that I'm showing you some other photographs of the Ireland outcrops that we worked. Thank you very much Yes, thank you for presentation. I think it's very very important work And I'm very interested and in this work as well What I got a little lost on was the uncertainty You know, this is something I very much appreciate as well It seems you're working with three things tank experiment process model and then Outcrops what Michael purge does is more process mimicking model. It's not really a process model. It's actually a statistical model So I'm a little bit worried about the horse in the carriage. What sort of comes first is that? How are you going to compare a statistical model? Which requires statistics and where would they come from they would come from your outcrop I've also got a question about scale compared to comparison of scales. You're looking at things that are at very different scales so and So maybe a few comments on that would be would be great Maybe you can have a talk so with regard to uncertainty I'm going to use the same answer that our first speaker use which is that I will There's a lot of uncertainty in what what we do here and I won't be able to necessarily address that in terms of Comparising Comparing our measurements from a natural system to what Michael purge does it indeed what he has is a rules-based model And what we can do is look at measurements from a natural system to test if the assumptions that were divide designing the rules-based model Yield similar results. So in other words if their assumption is that all sediment transport systems Always go to the absolute low that might yield a compensation index It's a little bit different than what we saw there because you'll note in all the systems that we saw there was a lot of Aggregation of the stratigraphy meaning that it was kind of staying focused So I think it's a method by which we could refine assumptions for rules-based modeling and as far as the process models again I think I would say that this is a place that we should be talking that Do that do the process do the outputs of process models match that of natural systems? So I'm kind of punting your question But the idea is that though I'll crop my perspective as a field geologist is very empirical I just have to correct what event-based modeling is for it's not for this It's a it's designed to take criteria and use it at a reservoir scale. It's not for this stuff Well, I would argue that what we did was work at this reservoir scale, but we're we use other things I'm a groundwater hydrologist and so and started out as a geology undergrad and My goal was always to use the geology to constrain the groundwater models just like you're you're being Paid to do this by Chevron, and I'm sure what they're hoping right and They're hoping for to be able to take this and constrain and understand oil flow in the subsurface No, so I was wondering if you had any thoughts about the utility Utility of your analysis to achieve those kinds of goals a more predictive model in oil or water Flow in the subsurface Well, you know one thing that we're working on is the more that we know about natural systems For example compensation stacking is one important way that you can juxtapose The sand rich parts of a channel to the sandwich parts of another channel if it's not very Compensational so if we see that a system is kind of sticky that means that the sand and sand contacts are enhanced and they're There's and also there would be perhaps better connectivity But then again, this is not what we're doing is we're being very empirical what I would argue is that if you're going to make a Reservoir model or a hydrological model of some sort that if your model matches that of what we're seeing in natural systems Then you can run the fluids through it and see how that and populate it with your permeability and prosody fields That you'll see how much of an imprint this really has and that's you know We're in the early stages of this obviously, but there are certainly applications. Yeah, really nice to see