 Welcome everybody, Lynn McCready set this all up for us and we're really honored to have systems let us present a webinar. I'm Chris Sherwood, I'm an oceanographer at the U.S. Geological Survey in Woods Hole and I'm one of the large group of people who've been working on the sediment transport components in ROMS in the coast model that John Warner has set up and I'm going to let Courtney introduce herself and then I'm going to kind of do the first part of the seminar talking about kind of the mechanics of what sediment transport is in coast and then Courtney's going to talk about the real science that she and her students have been doing with that. So anyway, go ahead Courtney, why don't you introduce yourself. Okay, thank you. I'm Courtney Harris, I'm a professor at the Virginia Institute of Marine Science and this is Danielle Tarbley. Danielle is a PhD candidate here at VIMS and so to show you some of the applications that we've been doing with the cohesive sediment model, we're going to show you some of Danielle's dissertation results and I'm also the chair of the CSDMS marine working group and CSDMS has asked each of the working groups to provide a webinar this year and for the marine working group, we thought that the webinar would be a great chance to show you all some of the recent model development that Chris and others have been doing to add cohesive transport processes to the ROMS-based community sediment transport modeling system. Okay, thanks Courtney. So I want to also make a call to Alfredo Zabaleta who is who helped me develop some of the flock components in the cohesive sediment stuff in coast and who he and I have been working on this presentation a little bit for the coast seminars when John Warner hosts his training for coast. Some of this has been taken from the sediment transport component and also both Courtney and I wanted to kind of keep this informal so you can send a chat and Lynn will unmute you and ask questions all along the way. Don't feel shy to stop us if something's confusing. So now change slides. Okay, so we're talking about one of the components of the coast sediment transport modeling system. The coast is a coupled atmosphere ocean wave sediment transport modeling system. It's been kind of designed and maintained by John Warner here at the USGS but it's got contributions from people all over the world and it's got a huge user base. We're mostly going to be talking about the sediment component but the other main components that are important to know is the ocean component which is based on the Rutgers originally the Rutgers regional ocean modeling system and we use swan for our wave modeling system although you can also use wave watch three and the atmospheric system is contributed by the weather research forecast model which is worth. There's also an infragravity wave component and a sea ice component and there's along with the distribution there's a bunch of modeling tools like MATLAB files and things like that and all this combined is the coast sediment transport modeling system. So I'm going to talk briefly about all of the sediment transport components not just the cohesive ones so that there's some context for understanding where the cohesive components came from and the things that are kind of related to sediment transport in the model. Obviously there's routines for eroding sediment, depositing sediment, changing the bed model so you can look at stratigraphy and today we'll talk about doing that for both sand which is non-cohesive or cohesive which is mud or mixed sediments. Settling is an important component in ROMS. One aspect of the way that ROMS has decided to do settling is that once a sediment class has a settling velocity that settling velocity stays fixed for the entire modeling run. That allows us to do very efficient and effective settling calculations but it has the disadvantage of not allowing us to change the settling velocity of some component kind of in real time as the model is running so that's just a feature that you have to keep in mind. The model includes bed load sediment transport using a couple of different equations and flux divergence so the calculation of erosion and deposition which allows us to calculate morphological evolution which is the aggregation of the bed or erosion of the bed. One of the new components that we put in for cohesive sediments is a flocculation and aggregation component and we do not explicitly include the sediment stratification correction in as part of the sediment boundary routines because it's included in the model. If you turn it on, the model turbulence closure equations will damp out turbulence if there's sediment density stratification so that's built in part of the model. An important part of any sediment transport scheme is that you have positive definite convection scheme. There are two of them now in ROMS and SWAN but you have to remember to turn those on if you're doing sediment transport because otherwise you won't be able to make sure that you conserve sediment which is one of the components we're shooting for. Lots of sediment transport equations need waves especially coastal ones. You can either specify waves as input or you can couple with SWAN or WaveWatch 3 and it can be either one way coupling where you just get wave information or two way coupling where you can send back changes in the water elevation that might be driven by the radiation stress from the back to the wave model so that it's two way coupled. We have wave current combined bottom stresses via a few formulations and the model has wetting and drying. The standard sediment transport components are fairly standard. One of the most important features of any sediment transport component is making sediment available for picking up. Since we have the layers in the model that compose stratigraphy there are, we have an active layer thickness based on the Harrison-Weyberg formulation that basically makes some of that sediment available and that active thickness is dependent on the excess shear stress. Erosion is a flux formulation where there are basically two parameters. There's an erosion rate which is E and a critical shear stress for erosion which has to be exceeded before sediment will come out and that is applied on a class by class basis. If you have several different sediment classes the critical shear stress and the erosion rate parameter may vary for each one. Sedimentation back into the bed is a flux formulation where it's basically the sediment settling velocity times the sediment concentration at the near the bottom of the bed and that deposition is just the sum of those two processes. There's a number of different bed load formulations. There's the kind of classic Meyer Peter Muir equation. There's a Solesby and Damgard equation that brings in a little bit of velocity asymmetry and allows and allows you to to somewhat correct for asymmetrical sediment transport. Both of those depend on the combination of wave and current components of the sediment of the bed shear stress. There's a bed slope effect so that it's harder to push sediment up a bed or down a bed and one of the things that we're working on now is we're working on adding another asymmetrical wave driven bed load transport basically on the Santos formulation and using bulk wave parameters to assess the asymmetry. So these are the sediment transport equations that have been in ROMs for quite a while and have been used for lots of successful simulations. Bottom boundary layer is for the most part when we're doing sediment transport we use the law of the wall so that the roughness is characterized by a z-knot, a bed roughness. That z-knot can be a combination of a number of things the kind of the Nicarazi grain size roughness, the z-knot for sediment transport, and maybe another z-knot for bed forms and those can kind of can be added. All these things can be controlled and the model switches you can turn on which ones you want to include and which ones you don't want to include. The wave current bottom boundary layer is important because when you have waves the orbital motions of waves increases the turbulence right near the bed and that couples the overlying flow to the to the bed more tightly and that kind of results in an apparent increased drag of the on the mean flow well and and that's characterized in in these formulations as a z-knot apparent which is usually higher than the regular z-knot and associated and there are a number of different wave current boundary layer formulations in here. We tend to default to the to the Madsen 1994 formulation when we do our model ones. When there are ripples on the seafloor that increases the roughness of the seabed and there are some formulations that relate the ripple roughness to basically the height times the steepness of the bed forms. Some of this is really form drag and doesn't come in as skin friction so we use the claim formulation to the smith and the claim formulation to partition the that drag so that we don't include the form drag component in the skin friction component of the bed roughness. So these are things are all turned on and turned off with if death statements at compilation time you basically set up the model compilation and so this is kind of a pseudo code over here on the right that shows the sediment flow and if sediment is defined then all these things occur if bed load is defined and you turn on bed load a suspended load is defined and that's included and even all of these have sub options and although we can't really get into that in great detail now we have lots of examples in the that John has included in the in the code where you can look and see how we turn on different components of this code. So what we're going to talk about today mostly is the is the cohesive component of the sediment transport and that includes both cohesive and mixed sand and mud components. We're going to talk a little bit about biodiffusion which allows the sediments in the bed to mix and change the stratigraphy we're going to talk about flocks and one of the things that having these components in the in the model while you do is to actually characterize the biogeochemistry of the bed which we're not going to talk about that much today but Courtney has been taking a lot of the a lot of these components that we've been working on and along with her students have been applying them to real-world situations so recently she's had a number of students who have taken these components of ROMs and published great papers on them JP, Kelsey, Julia and Danielle are all taking have taken components and and done really cool science with all of them. So here's a quick example of how the non-cohesive bed works this is one of the first simulations that John Warner did in Massachusetts Bay he basically applied a series of typical nor'easter storms like the one we had all weekend and he started off with the kind of a uniform sediment distribution on the bed and and he did several times with these storms so here's a map in the upper left hand corner of the bed stress kind of at the peak of one of these storms and the higher colors or higher bed stresses typically in the shallower areas over on the right is the bath and bathymetry change that's the biggest changes here are only five millimeters but it shows that there's erosion in some of the places with high bed stress and a little bit of deposition in some of the quieter deeper basins and on the left is a map of map of sample grain sizes in the area and on the right is what what the ROMS surficial sediment ended up looking like after several these storms hit and this kind of gives us some kind of reassurance that the model is changing the bed sediment in kind of reasonable ways using both the stratigraphy and the sediment transport routines we've got so that's non-cohesive sediment let's talk about cohesive the cohesive the cohesive bed model is is built on a very simple and and and somewhat heuristic idea that there's an equilibrium well first of all that that in cohesive sediment rather than worrying about the individual grain size critical shear stress what's important is kind of a bulk critical shear stress of the whole bed so the idea is that if it's muddy and cohesive you can't erode one grain out of that bed at a time you basically have to exceed the critical the bulk critical shear stress of the bed in order to get erosion and of course individual grains come out but but their erosion rate is not dependent on their individual critical shear stress and that in general that profile of critical shear stress changes with depth in the bed typically it increases as the bed is increasingly compacted and so we we have this concept of a equilibrium critical shear stress profile in the bed which typically increases with depth up to some asymptotic value so the curve kind of looks like the one on the left and the idea is that if you erode that bed you kind of cut into that profile and truncate it so you end up with something like this profile over here on the right where the critical shear stress is now higher up at the bed top of the bed if you deposit a bunch of fluffy non-cohesive sediment on the bed you may end up with a big pile of kind of less easier to move less cohesive material on the top of the bed and presumably over time that will compact and you'll work towards the critical towards the equilibrium profile alternative is if you've if you've truncated that profile and exposed difficult to move material to this seabed that that will kind of water and swell and gradually maybe bioturbate gradually go back to the equilibrium profile the concept is that that takes a lot longer to happen than compaction which happens fairly reasonably so we think there's something like an order of 100 times difference between the time scales of those two processes but the idea is whenever you put the bed out of equilibrium it'll try and get back into that equilibrium at a certain speed so here's an example of what happens during erosion the initial bed level was right here at the z level and then we applied a critical shear stress that was in excess of the critical shear stress at the bed and we truncated the bed we eroded about two millimeters of the bed now the critical shear stress follows this follows this black curve and and it shows you that the material right at the surface is difficult to erode probably the erosion process stopped as as that material became harder and harder erode and that's what that's what caused it to stop eroding when it got down by two millimeters the the equilibrium curve is in red and so over time now if the if nothing else changes that will slowly migrate back towards the black curve i'm sorry i have this i'm sorry the black curve will slowly migrate back towards the equilibrium red curve the opposite happens if you're depositing on a cohesive bed in this case on the left the bed level was at zero water was up here in the blue we deposited two millimeters of material and the critical shear stress of that material the equilibrium shear stress is shown in red and you can see that that material at the surface is easy to move but over time as it compacts that material the black curve moves towards the red curve and it becomes more difficult to move as it compacts so that's kind of the idea here's an example of the model being run doing that and i won't go over this it's basically the same sequence of events the erosion rate truncates the bed and moves sediment after a while that that sediment swells and becomes easier to to move and then there's a deposition when the shear stress time series decreases there's deposition of fine material at the top and there are a number of different parameters that control this that that we need to worry about but i'm going to focus more on examples here over on the right is a is an example that includes the stratigraphy of the bed and what we've got of sand excuse me several size classes yellow is sand and the darker colors are mud this top panel shows how much what materials in suspension of the bed the middle panel is a kind of an artificial time series of bed shear stress where we start with no shear stress we have a high stress event which then stops low shear stress another high stress event with kind of a different shape and then no stress and the active layer thickness that's associated with these and changes as the stress changes is shown in the bottom panel and so what you're seeing is that you apply a large amount of shear stress and then when you decrease that shear stress at the end of this event materials start settling out so that the concentrations are decreasing in the in the water column and you notice that the sand falls out first as it should because it's got a higher settling velocity and then the silt falls out or the finer sand and then the classes of mud and so over on the right you can see that this stratigraphy that was built after this resuspension and erosion resuspension and deposition event occurred is a typical finding upward sequence where you have high amounts of sand deeper in the sediment column and increasing finer sediment as you go up and so this is another example where it's kind of a simple conceptual model that produces what we would expect from our knowledge of how sediment is supposed to perform in the how sediment is supposed to behave in the water column so the mixed mud for the mixed bed behavior is an attempt to combine the two end members we talked a little bit about the non-cohesive sediment where the erosion rate depends only on the particle critical shear stress we talked a little bit about the cohesive sediment bed where the critical shear stress depends on a bulk critical shear stress that applies to all the sediment and the idea of the mixed bed is that there's some ratio of sand and mud that produces those two end members so if you've got more than about 20 or 30 percent mud you're going to behave completely cohesively and so it's going to behave as a cohesive bed if you're if you're if you've got a very small amount of mud maybe less than something like three to ten percent there's no cohesion at all and the mud is just in the matrix of the sand and can be resuspended by the sand and then in the middle for lack of any other better approach we basically have gone kind of a linear combination and what we're doing is we're saying that the critical shear stress for resuspension kind of gradually goes from one end member to the other as you go through as you mix these sediments and and so what we are changing as we go through that as we as we look at the the non-cohesive component we just call this p-coe we switch from being totally non-cohesive to totally cohesive kind of linearly and that produces kind of recent reasonable results here's the flux of clay based on how much mud is in there so if you have very little mud you're over here on the left left silt and clay can be easily resuspended from the matrix during a storm and so the erosion flux is very high it kind of gradually decreases and then when you get over under the right you're behaving completely cohesively and that material all comes out together whenever the cohesive shear stress is eroded unless unless the material is so difficult to erode let's say you've got pebbles in a mud matrix or something like that those pebbles won't move and come out of the bed unless not only the matrix that the bulk cohesive stress is exceeded but also the critical shear stress for those individual pedals will be exceeded so so it's a it's a kind of a heuristic overall approach but it seems to produce kind of reasonable results um finally one of the component that's important um for for any model that has stratigraphy in it is some kind of mixing within the bed because in the real world there's often either ripples migrating across the surface which tend to mix down to a certain level or there are animals fauna in fauna in the bed which move the sediment around and this schematic over here shows the importance for geochemical processes often sediment is is resuspended into the water column and then it can react with components in the water column with either resuspension and desorption of say contaminants that are in that sediment or adsorption of something that's in the water column onto the particles and then they deposit down and and they form a layer that that is kind of uniform initially at the top but then it mixes and mixes that bed in and we kind of envision that there's some kind of a profile that looks kind of like this you can you can make a profile that includes any of these components as input to the to the model the idea is that near the surface there's kind of a high mixing rate and it might be uniform for a certain thickness and then way down deep some place there's no mixing associated with in fauna or surface there that there may be molecular if you see the in the pore water or something but essentially there's zero mechanical mixing and then we kind of put an exponential profile in between those two end members so that's kind of the conceptual model that we are using for our mixing and when you combine that with the mixed bed model you get a number of different results depending on where you are and and if you this this is in the paper where we describe it and I won't go through it in detail but in in you can see that the stratigraphy that results from changing the from non-cohesive to mixed bed and from changing the bio diffusion varies and and in this case with the large bio diffusion the bio diffusion mixes down to some bed with the mixed bed and so you're getting changes deeper into the bed with this case then you are with the small bio diffusivity diffusivity which doesn't reach as far into the bed and so so there's a kind of a tapering of the mixing in the ultimate stratigraphy when these are part of some of the kind of the test cases that there were conceptual cases so that we can make sure that the model was doing things that seemed reasonable finally one of the important things that we've got in the model is the flocculation component and and we're following a model called FlocMold that was developed by Vernet and and he has helped us implement it in ROMs and the con the idea is that there's a fractal dimension that relates to the density of the flocs with the volume of the flocs and ideally if the fractal dimension is three the flocs are completely solid so they're regular spheres and their and their flock diameter is the same as their solid spherical diameter so there's no no difference between these two components on the right but as you reduce the fractal dimension the density of the flocs becomes lower and the flock size becomes relatively bigger for for an equivalent diameter and and so we relate the size to the to the fractal dimension and the change in the in the grain size of any particular flock class is a function of gain by aggregation gain in that class by break up by shear of a larger class so some material comes into this break up by collision of a larger class which might put some components into this lost due to aggregation so two sizes in the size class might aggregate and so that material will be lost to a larger size class and lost by breaking up from this size class and lost by breaking by collision the shear and the collision are both in those rates are both increased as concentration and turbulence increases as is so so if you have lots of turbulence you can break up flocs aggregation requires both turbulence of particles see each other but also the higher the concentration the more likely the two particles will see each other and and aggregate so a typical typically we will break our class sizes into a number of different break our sediment into a number of different size classes and distribute them logarithmically by size and then the settling velocity is associated with the particle density and changes as a function of size and and the mass basically is moved among these classes through aggregation and and disaggregation depending primarily on the shear stress and the density of material that's there so we did some test cases this is a a test case where you've got a kind of a tank of water and you increase the shear the concentration you keep the same amount of sediment and increase the shear in that tank of water and the floc sizes decrease in as the shear rate increases as they get torn apart then you decrease the shear the shear rate is the gray pyramids here you decrease the shear rate and the flocs are still being moved around and they aggregate and then they end up with a larger diameter in the real world that would settle out but in this tank they can't really settle they can settle to the bottom but they get re-suspended and this is one dataset that we can use to demonstrate that the model it seems to be doing the same thing that the field experiment does another idea is whether or not the flocs come into kind of some equilibrium diameter in the in Winterwerp's original one of his original papers he argued that there's a linear relationship between the floc size and the ratio of concentration in the square root of the shear in the water column and these linear relationships change depending on the aggregation and the breakup parameters so each one of these curves represents different parameters these are these are somewhat tunable parameters that determine what the equilibrium floc size is depending on the rate ratio of concentration in shear and so these are model results in the dots in blue and linear fits that show that the model does kind of come to equilibrium sizes depending on the conditions and this is kind of a very simple experiment where we've where we've taken the water column and just taken it from a still water column with initially uniform distribution of flocs in the water column so this is the floc concentration in the pop panel the elevation the mean diameter the weighted average diameter of the flocs in the in this column and the settling velocity of the flocs in this column initially the flocs are uniformly distributed the water column is accelerated the flow velocity increases the turbulence increases initially the flocs start settling out but as they settle out they end up in the high concentration region in the bottom where there's a lot of shear and a lot of concentration so there's a lot of aggregation and over time this whole system kind of comes to a dynamic equilibrium where you have a concentration profile that looks like this you have a diameter which looks like this so you have smaller particles up near the top of the water column larger particles near the bottom and you have an average settling velocity that looks like this and this little kickback in the right near the bed is is actually the wave current bottom boundary layer where there's so much shear that the floc size is cut right down and these profiles for each individual class are shown over on the right so one last thing I wanted to mention before Courtney talks about how we actually could use this model how she and her students have used this model is that there's also a vegetation drag component now in in ROMS that was put in by Alexi Baudin and Neil Gange and others here in this lab and the idea is to be able to put in for ag associated with submerged aquatic vegetation vegetation and so there's an interaction between the parameters from the swan model and the in the flow model that go through the vegetation depending on different parameters and it has a number of effects it it within the canopy it reduces the amount of tke and so there's less mixing in the canopy but above the canopy the the flexible component of the submerged canopy can enhance the tke and so the vertical distribution of particles that are in the water column and are quite different if you include the SAV component of the model so here's kind of a summary of the different components of that model and I recommend that that you have a look at the paper if you want to include that in your in your model but it's for modeling for example marshes or seagrass beds this seems to be a very effective addition of the model so I'm about to turn it over to Courtney and I'm going to let her start sharing her screen the way all this is accomplished in the model our details that are probably too difficult to get into here but basically you need to include changes in the .h file which changes how the model is actually set up you need to make changes in the input files that describe what kind of input you're providing the model you need to have you need to describe the sediment components that you've got and you need to either initialize the model analytically or with some kind of initial file that shows the spatial distribution of all these components in there and there are good examples in the ROMs distribution of how to do this but we're also welcome to answer questions here around the forum as to how to actually implement this and so I think that's it for me Courtney maybe I should stop and ask if there are any questions while Courtney shares her screen and I'm going to unsharing I'm going to stop sharing mine yeah so if there are any questions while I figure out how to share my screen that would be great I notice that Aaron Beavers on there and I forgot to include him as one of your students who is doing very cool stuff with this model so sorry about that Aaron and all right are you guys Chris yeah this is Larry Sanford hi hi Larry so how when you when you have bioturbation in the model do you are there any issues with numerical diffusion between the layers there's a little bit of the numerical diffusion but it's much much smaller we're using a we tested that kind of extensively and we have there's there's more diffusion than there might be because there's variable bed thickness brought into there so there's a there's a dz component in the in the equation but we're we're using a fully implicit solution that is stable and has errors that are less than one percent the numerical diffusion is less than one percent when we do a bunch of analytical tests with analytical distributions okay thank you okay um are there any other questions before we move on to trying to show a few applications of the cohesive model okay um so what I want to do with the kind of next 10 minutes or so is to show you some of Danielle Charlie's PhD results so Danielle is a PhD candidate she's co-advised by Carl Friedrichs and myself so Carl is co-author on all of these the work she's doing and Chris is also on her dissertation committee so he's also been heavily involved in the model applications that she's been doing so one focus of Danielle's work is to try to quantify how the cohesive sediment transport processes impact the York River estuary which is the uh system that is you know right outside my window here at VIMS so she's been using the sediment transport model that Chris just gave a great overview of and she's been specifically in not including the non-cohesive version which Chris covered in the beginning of his talk but accounting for the cohesive version of the model and specifically looking at the effects of flocculation and bed consolidation so here um this is showing kind of schematics from what Chris already covered Danielle's model runs include both bed consolidation which uses the model that Chris covered that was originally developed by Larry Sanford and it includes this kind of parameterized critical shear stress at equilibrium depth profile and then the model tracks the instantaneous critical shear stress for erosion and nudges it at each time step back towards the equilibrium value and then on the bottom here is just a schematic of the flock model that Chris just discussed and Danielle has been using this with I think about 11 size classes to try to account for flocculation processes in a system like the York River so just a shout out to the work that this talk was kind of put together based on stuff that Danielle presented just last week at the surf meeting in Alabama so for her model runs she has set up an idealized estuary model and she based it the kind of overall dimensions and hydrodynamics of this model on the York River estuary but one thing to note is that we're using this idealized kind of 2d representation of the York because especially the flocculation model was computationally expensive so we wanted to get a feel for how important the flocculation processes were in this idealized system before we attempted to run it in a full 3d model which would have you know many more grid cells so the basic features of her idealized model grid is we have a riverine input I'm not sure if I don't think you guys can see my pointer we have a river input to the right coming in relatively shadow shallow water putting in fresh water and then on the left hand side the downstream boundary condition is a tidal boundary condition with salty water flexing in and out with the with the tide and then underneath the water column is a sediment bed model that accounts for the bed consolidation processes that we talked about that Chris talked about earlier so we're going to show you an animation of the model run so the top she ran the model for this case twice once she included the flocculation processes and then the second time she ran it she did not include flocculation so here the arrows will be the current velocity the red lines represent the isopignals so showing you how the salt distribution is set up and then the color is suspended sediment constant concentration in milligrams per liter so now if you can see the time dependency of the model we see the tides flexing in and out and we see that the sediment concentrations are heavily weighted towards having high sediment concentrations here in this turbidity maximum region and we also see that there's quite a big difference in overall suspended sediment concentration when flocculation is included compared to when it was not included so again the major features that we see is the difference in sediment concentrations when we include flocculation and that whether or not we included flocculation we did get an estrange turbidity maximum that's formed by the sediment trapping at the halt at the head of the salty region so in this model in her 2019 paper Danielle ran the model with and without each of these cohesive processes so on the bottom panel here on the on the right is showing the overall depth integrated amount of suspended sediment concentration for a model run that that included both consolidation and flocculation and so that is the blue line and then in the green and the red lines those were models that either for the red line did not include flocculation and for the green line did not include bed consolidation so with both of those we see that we see two kind of twofold differences in overall suspended sediment concentration when we neglect those cohesive processes and then here she differenced the suspended sediment concentration in each of those in each of kind of the modeled size classes I think she included 11 different size classes ranging from one micron up to a thousand micron sediment wherever you see blue that means that when she included flocculation there was less of that sediment in suspension where you see red that indicates that when she included flocculation there was more of that sediment in suspension so what we see is that our main differences in sediment concentration are in the etm and that flocculation there acted to aggregate the finer sediment and package it repackage it as the coarser aggregates and then something she's been looking at more recently is to try to evaluate how good this equilibrium model is when we apply it to our idealized estuary so chris already showed this figure where he ran the flock model to steady state for three different cases and showed that at steady state it did meet winter ropes ideas about equilibrium flock sizes um danielle pulled out times from her idealized model where she thought that the model would likely be at steady state at this equilibrium flock size and so that was at i believe the peak flood and ebb times for the etm locations so each dot represents the equilibrium uh or the the instantaneous size that was calculated by the model and then the the black line on it is the same equilibrium line that she obtained the black line represents a best fit to these points and then the color of the dot indicates how high in the water column that point was so the dark blue dots are the very near bed data points and then the red ones would be in the upper water column so overall we conclude that during these times of peak flow that the the model is pretty much approaching an equilibrium flock size distribution but when we plot all of the scatter for the instantaneous diameter from the model compared to that equilibrium flock scaling shown as a black line we see that a lot of times the model is producing flock sizes that are very different from the equilibrium flock model so here the very bottom grid layer is shown as these black triangles and then the colors of all the other dots indicate how high up in the water column it was with blue being near the bed and red being near the water surface so we see that throughout the water column there are times when the model does not meet this equilibrium size distribution so what she's been looking at in her analysis more recently is trying to evaluate the importance of flocculation throughout different time periods and at different locations in the in the in the idealized model so I'm just going to show a few of her recent results kind of and these point out the importance of the flocculation terms as well as when the model is at this equilibrium flock size distribution and when it is really not quite there so on these panels the the very right hand panel oops sorry the very right hand panel shows the time the kind of title where we are in terms of the tide in the model so so we're showing that we're going to show you results from points one two and three which represent slack tide um raising flood tide and then peak flood tide and then the panels to the left show model results from each of those three time periods on each of these panels we're showing the grain size diameter with the finest stuff being to the left the course of stuffing to the right and we're showing the the mass the amount of mass that's being transferred through flocculation for each of those size classes so wherever these bar graphs are above zero that means that that size is being added to by the flocculation process and wherever that bar graph is below zero that means that the flocculation process is acting to take material out of that size class so here during the flood conditions at times two and three we see that we're disaggregating our larger particles and form and they as they break up they're forming more of the small of the medium size flocks and then the other data points that are in these graphs are these dashed vertical lines the red dash vertical line is the equilibrium diameter and then the black vertical line is the instantaneous modeled flock diameter so we see that during times two and three the equilibrium diameters are finer than the instantaneous and so in order to nudge the system back towards equilibrium what the model is doing is breaking up the larger flocks to form those medium size flocks okay and then I should have pointed out that this model data is from the very near bed so at the very near bed at peak flood we're not at equilibrium we're resuspending stuff that is coarser than the equilibrium flock size and that stuff is quickly being disaggregated and put up and then put into the finer sizes and then these panels are the same panels again looking at conditions as we go from peak flood down to peak ebb and again we see pretty similar behavior with with we tend to resuspend sediment that is coarser than the equilibrium and when we are at higher shear stresses that stuff breaks up to form the medium size flocks but that was all very close to the bed when we look at what's going on about a meter above the bed so for these results daniel took model estimates from a grid cell that was 90 centimeters above the bed and there we see that the system tends to be reaching more of the equilibrium sizes when we are more at the peak flood because the equilibrium flock size diameters closer to the instantaneous one at times at time three when we're at peak flood and they're also in this upper water column we've switched from disaggregation being important to aggregation being important for most of our mass balance terms we have we're losing sediment mass in these medium sizes and that stuff is especially at peak flow conditions that stuff is aggregating to form the coarser the bigger aggregates so to conclude some of the results that we're seeing from this idealized estuary case is that the idealized estuary case does reproduce the key features of the of our idealized estuary such as estuarine circulation forming the etm but in order to get realistic values for the sediment concentrations daniel needed to include the cohesive processes of bed consolidation and flocculation the flocculation had the biggest impact on the suspended sediment concentration in the etm and it reduced sediment concentration by about 50 there outside of the etm the bed consolidation had a bigger effect and it but it also decreased sediment concentrations by about 50 and then in daniel's more recent results we're seeing that flocculation is transferring as much or even more sediment in the flocc term compared to the horizontal and vertical advection terms i didn't show those results today just for time but that's kind of one of the significant results that we're seeing out of her model runs and then finally that the model that accounts for this time dependency in the flocc size showed that in our idealized estuary a lot of times the flocc size did not reach the equilibrium flocc size it seems to only really reach the equilibrium flocc size when we have peak flow conditions and up above that immediate resuspension layer so that's what i had if there are questions either for me or chris or daniel and i presented this today it's really daniel's work but she's been um busy writing up her results so she i'm sure so maybe if there are no i can't see the chat thing now lin was going to keep an eye on that to see if there are any questions it looks like uh you may have two questions coming okay i think somebody's probably typing at this moment daniel's trying to help me figure out how to do this okay okay so there are two questions okay so daniel the first question is why didn't you present this because i'm trying to write it up in word form so did you spend the weekend writing or working on this talk i spent most of the weekend writing she did present to that surf um a week and a half ago and did a much better job than i did today probably so and did you have another question have we thought about the effect of hyperpikinal flows yes we have thought about it um and there are examples of the roms model being being used to account for hyperpikinal flows shenan chen did a kind of similar to our idealized estuary shenan chen did a paper where he used the roms model in a 2d across shelf case to account for hyperpikinal flows um erin beaver who i believe is on the call as part of his dissertation he looked he included that um the sediment density effect to look at whether um a river in new zealand would plunge during high um discharge um the shelf off of that particular river right off of the river mouth is not very steep so that river wouldn't really um plunge too very deep because it kind of emptied into an embayment and so the so i guess the way i would include hyperpikinal flows from a river mouth right now would be to try to have sufficient vertical resolution so that we would have some hope of capturing the suspended sediment stratification at the kind of top of that hyperpikinal layer and but to date i i have not seen cases where people ran it for kind of a realistic um um model grid and bathymetry so maybe larry's question so when you're using the flock model you've got christ said that for you have a fixed number of sediment classes and each class has a fixed settling velocity right correct did that right so did you set your so you all had one basic fine sediment size and you just divided your sediment classes into different classes of flocks is that the way you did this and each class of flock had its own settling velocity yes oh okay and it assumes that all classes of flocks are have the same fractal dimension right right so the settling velocities are just what what's assigned to it by the same fundamental size and the same fractal dimension within their different categories of flock okay cool and also do you have what's what you have the the actual citation for your 2019 paper yeah i think it's on Courtney's presentation but it's in the journal marine science and engineering okay thanks i believe the next question was from how wing um what would like to know how our group creates rom's grids and what we use for grid generation christ have you done that recently i my group has done it and i used an m file that was called easy grid and i decided that might be a misnomer but we tried to update easy grid that's what we did and and recently i've been doing it um in python but it's not it it's been very um it hasn't been a gooey operation it's it's been you know actually me picking the points that i want to do and then doing the smoothing and the and and all the things for the the symmetry grid and then finally writing out of rom's grid so there's uh there are some getting getting from an xyz grid in matlab to a rom's grid is pretty easy because there's some tools in the rom's toolbox for doing that but slapping down a grid a curvilinear grid on a on a you know a pythymetry map um there are a number of ways to do that and i don't think any of them are completely satisfactory but that but um easy grid's probably the best way to start i remember at the coast user group meeting last february that they asked for a show of hands of who had used easy grid and it was like me and one other person so and john and danielle's reminding me that john warner had some other tools that come with the coast distribution i think so so i would probably look at those first before easy grid or chris's python tools that he mentioned okay there was a question about whether the bed consolidation module changes the sediment bulk density to reduce the layers thickness within the bed layers so so the answer is yes it does do that so the mass of the sediment is conserved the bulk density changes and the thickness of the layers change but none of that those things aren't really calculated dynamically they're only calculated kind of um uh through this equilibrium profile approach okay yeah all right um and chris are you following the chat so brian romans has a question um uh applications to the deep sea three to five kilometers water depth contra right currents and drifts um so and brian is wondering if we've used cohesive models in that context so i have i've run the roms model to try to represent the gulf of mexico continental shelf to slope and we do get in the model contour currents we do get resuspension from fast relatively fast flowing currents along the continental slope so in the model the shelf is definitely wave driven transport dominates but when we get deeper you know in the gulf of mexico off of when we get beyond the the shelf slope break what dominates resuspension is fast currents that come through now whether those whether those are real we don't really have good data time series data to evaluate how good those contour currents that we get in the model are and that is something that i think is important to try to do is a more tightly coupled deep sea model data comparison and in my model we i have not used the cohesive model at at this point yet we've used a non cohesive formulation and i can't speak to what other people have done in the deep sea but i think it's important to try to get better studies howing wrote on the chat to try that he might try easy grid he's tried grid builder and not liked it he or she okay are there other questions or any that we missed in the chat no well thanks to everybody for participating and thanks to lin and systems for organizing this