 do a little bit of logistics before we get into the overview of ROMs and then we're going to have some hands-on ROMs activities. So the way that I'm formatting this clinic, we have about a three-hour time frame, but in the middle of that three hours, there's a half an hour or 15-minute break to get some coffee, which I'm sure you'll need after listening to me for an hour-and-a-half. And then when we come back from the coffee break, I'd like to have a discussion to get some feedback on how you think this hands-on exercise and the idealized model that I'm showing you is going. And part of the motivation for doing this clinic was that there's been a call in the Marine Working Group and the Coastal Working Group to have some idealized hydrodynamic models within systems. So one motivation for setting up this idealized model that I'm going to show today is that we might be able to provide this tool within the systems as a idealized shelf hydrodynamic model that could be used for whatever application people wanted. So once you get a chance to try it out and analyze some data and try running the model again, then I want to kind of brainstorm on where we should go forward with this type of effort. Is this the kind of thing that the working groups need or are there other approaches that would be better or are there improvements we could do to this model? So the timing will be I'm going to give an introduction to ROMs into the test case that we're running for this clinic and then we'll have time for you guys to kind of roll your sleeves up and analyze some of the model output data. We have done some groundwork so that at least a few of you should be able to run to make some changes to the model and rerun it. We can't all redo it. That would bog down the system. So I think when we get to that point we'll kind of make a vote of how many people have access to beach that could conceivably rerun the model and then maybe pare it down to just maybe one to three model runs that we start up as kind of a group effort and then after coffee break and after our discussion there will be time to analyze those model runs. I think it takes about 12 minutes to run the model for the time period that we've set up. So like I said the motivation, oh and so before I start, the way we've set up, the way we in our group analyze net CDF ROMs output is to use MATLAB. So we have on our thumb drives, on our thumb drives we have a few MATLAB scripts that you can use to analyze some of the ROMs output. So the hands on part of the clinic will be analyzing net CDF output. So I'd like to see a show of hands of people who cannot analyze, who don't have MATLAB and then maybe those of you who do not have MATLAB could partner up with someone who does. So raise your hand if you have MATLAB and you're willing to work with one of these other people that don't. And actually this clinic is based on a class that I teach and in the class I always pair the students up. For one thing we have about six computers in the lab so we need to kind of combine forces because of the number of computers we have but I find it works a lot better when people work together anyway. So once we get to the hands on part then those of you who don't have MATLAB if you want to browse the model output choose a buddy who does and you can work together with them. So I'm calling this clinic ROMs for the non-specialist and I kind of gave one of the buzzwords that came out of our all hands meeting last year was a call for something that we called ROMs Lite. So ROMs is a regional ocean modeling system and it's a pretty comprehensive community coastal ocean model but within the systems community there's a call for something that's not quite as complicated, not quite as many options to choose from, kind of pared down to more of the bare bones of what we need ROMs model and we called that ROMs Lite. So to prepare for the clinic I tried to set up something that could be a ROMs Lite that people could hopefully use with less of a learning curve than implementing the full ROMs model. So this is an example movie from one of our group's recent efforts in running a full ROMs model. So I guess I should have introduced before I started Julia Moriarty is a graduate student at VIMS and Tara Niskern is a research scientist at VIMS. I want to thank them for helping me put this clinic together and they've done a lot of work here this week and ground work to get us to this point to have this clinic. So I appreciate y'all's help and to start the clinic I wanted to show you an example of kind of the full, the ROMs heavy maybe we would call it. So this is what Julia's model run that she produced for her master's thesis and she recently published it as an Estren and Coastal Modeling Conference paper. So there's a citation for 2014. So in this model run Julia developed a model grid for this site on the North Island of New Zealand. It has a river input into the embayment there and there are energetic waves and currents moving the sediment around so we've got a river coming in we've got energetic waves moving sediment she had to get her wave input field she had to nest it within a spatially variable wave field so that that was a bit complicated she had to nest her open boundaries into a larger scale oceanographic model to get the along shelf currents. So that was a bit added complication. She had to worry about cross shelf and depth stratification on the shelf so she had to nest that within a large model so it took her a while to get this model up and running a big effort and but once you go through that effort then you have this tool that gives you hopefully sediment fields that look like something in the real ocean but it takes a big commitment to developing that model. But there's been a lot of effort within there's been some effort within systems to to get some hydrodynamic models into the into the repositories and so I put this figure up here this is the chess ROMs model it's a community ROMs model of the Chesapeake Bay and so it's a three dimensional hydrodynamic model and within this a variety of different modules can be loaded a sediment transport model like we'll use today but also models for biology and other other components and because there was a big call from the marine working group in the coastal working group to have a hydrodynamic model within systems systems devoted quite a bit of time to putting chess ROMs into the repository so so it is available but I think the general feeling is that even though a lot of effort went into putting this model and other Chesapeake Bay models in the repository as well as putting a model that's a more general not a Chesapeake Bay model but a general ROMs model in the repository that it hasn't really been used that much so one thing I want us to think about when we come back for discussion is if there is a need for this kind of model what can we do to capitalize on the investment that's already been made to so that the resources would be used the framework for the clinic today is we're gonna I'm gonna overview the ROMs model framework and the IO structure of the model and then we're gonna look at an idealized continental shelf model that's called the River Plume 2 model in kind of ROMs jargon and then as participants you'll have a chance to look at model output and as a group or as a as a set of smaller groups will will make some changes to the idealized model and rerun it and then you can look at the model output from the modified test case and then after our coffee break I'd like us to to again brainstorm for how to go forward with this type of thing I took this slide from a talk from Hernan Orango and this is Hernan's kind of slide where he describes what's in ROMs the regional ocean modeling system so Hernan is one of the main developers of the ROMs code he's at Rutgers and here he's showing maps of different ROMs model output from around the globe and he has a list of the main features of the model so in general it's a free surface hydrostatic primitive equation model and it uses terrain following vertical coordinates I'm going to go I'm going to show you guys more about the vertical and the horizontal grid because that that is really relevant for for how you analyze the model output but in here he points out that ROMs has higher order numerical schemes and it's parallelized so it can be run in MPI so it can be run in a high-performance computing environment it has a lot of effort in ROMs has gone into data assimilation so that's all that about tangent linear represent or adjoint models so a big part of the ROMs community is interested in data assimilation but the community the group I've worked with more is is in this other bullet for tides ecosystem sediment and sea ice models within the ROMs hydrodynamic model there are ecosystem and sediment models that are linked to the hydrodynamics and so one power one thing that makes ROMs a nice environment to work within is that you can have a physical oceanographic model but you can also have access to to geochemical ecosystem and sediment transport models these are slides I made they don't have the pretty picture that Hernan had but the so the most relevant things I think for this group about ROMs is it's a 3d time dependent hydrodynamic model that solves the basic conservation equations it includes transport equations for temperature and salinity and those transport equations can also be applied for sediment tracers as well as biogeochemical tracers it's a community model right now there are over 4,000 registered users and then in addition to the hydrodynamics there are modules as I said for sediment transport and biogeochemistry so today we'll use a sediment transport model and that the model will use as described in Warner et al 2008 and it can account for bed load and suspended load but we'll just use the suspended load today other things about ROMs that make it nice for coastal applications is it can be run as a nested model so you can have higher resolution versions nested within the lower resolution there are versions of it that can be run with wetting and drying so you could if you you could set it up to do storm surge or run it in macro tidal environments and there's a big data assimilation community a lot of effort especially within the group that developed the sediment transport model John Warner and his colleagues a lot of effort has gone into coupling the model the hydrodynamic model with other models so a lot of the effort has gone into coupling it with swan and with wharf there's a lot of the development effort in ROMs has been in developing the data assimilation parts of the code that's what I meant oh not for sediment now that's that's a wide-open niche as far as I know no one has applied the data assimilation they've done it for temperature salinity currents and and the biogeochemists have done it to choose their parameters for their biological models but as far as I know the sediment community is not yeah so a lot of the data assimilation has been used to set open boundaries or physics or for the biological process so to continue our overview of ROMs it's written in Fortran 90 it's it's parallelized so in theory if you make changes to the code you should be able to recompile and run it in parallel it includes a lot of options so it includes different advection schemes that you can choose between it includes a range of turbulence closures and it includes a range of bottom boundary layer treatments so on the one hand that's really good it gives the modeling community a lot of flexibility to choose which of these options that particular application needs but on the other hand I think that's one of the things that makes a learning curve for ROMs kind of steep is that you you have the sense that you've got to choose between all these things and sometimes if you just want an idealized continental shelf model you might not want to have to think about these you might just want to get a model where reasonable decisions were made for these different options model output is in in net CDF and the but the model input is a is another place where the modeler has a lot of choice and so there's a flexible mix of model input if you have a long time series of model input like if you have a wind field with time and spatial variation in the wind field you'd want to put that in in a net CDF file but if you just wanted to run your model with a constant steady wind then you you don't want to generate a huge you know 5 gigabyte wind field to just have a constant steady wind so if you did that you'd you'd want to hardwire it in the code that would be the easiest way to set a steady wind field so ROMs gives you the flexibility to make that choice to do it the way that you think is easiest it might be some of the model input is typed into a text file some and then some you have a choice between hardwiring it or putting it in net CDF so again you have the flexibility but with that flexibility comes a bit of a learning curve because there have been times when we changed a model input like we changed the text while we change a model input and we look at the output and it didn't change and then we have to sort through and figure out that well it wasn't getting the input from where we thought it was getting it actually from hardwiring or from from a net CDF file so so with the flexibility comes a kind of a layer of complexity that gives makes the learning curve a little steeper for the for the grid any hydrodynamic model it's important that you think about what the grid is because that's one of the main differences between the different models that are out there ROMs uses an orthogonal curvilinear grid so it's a structured grid where the curvature of the grid can be chosen to kind of mimic the bathymetric curvature of the coastal area to the extent that you can still keep your grid cells pretty orthogonal and it's a staggered grid which most structured ocean much most structured grid ocean models are staggered grids but that the fact that it's a staggered grid for our from our point of view if we're not getting into the numerics but we want to look at analyzing model output we need to keep in mind what our horizontal grid is so it uses an arc how a sea grid so what that means is that the different model variables the velocities and the tracer concentrations are calculated at different locations in the grid so on that figure on the right each one of those squares represents a grid cell and then in the middle of each square there's a little green row that's called the row point of the model grid that's where the density is calculated and the bed stress is calculated and sediment concentration is calculated in the middle of the grid cell and then the velocity the red u's are the u points of the grid that's where the cross this might be the cross shelf velocities are calculated on the grid face not in the grid center then the purple v's are where the v velocities are calculated so there on the maybe that would be the along shelf faces of the grid so one thing I like about roms is in your output file it tells you exactly where the u points the row points and the v points are but what that means is that if you what it means in terms of when you get a model output field if you get a u field and a sediment concentration field those those the sizes of those data arrays are going to be different because because you have more u faces than you have middle of the grid cells because you have boundaries for each so your u you're going to have one extra velocity for each of your sediment concentration so your post-processing scripts just have to interpolate things to the basis if you want to calculate something like a sediment flux in terms of the vertical grid since we're running sediment transport then we have kind of two parts to our vertical grid in here this this is a figure from John Warner's paper and the blue grid cells represent the water column model so that's the normal hydrodynamic model the brown grid cells represent the sediment bed model so in the the model that we will run we're going to have 13 water column grid cells and they overlay 10 sediment bed layers okay so here the blue grid cells will have grid cell 1 up to n equals 13 and so we have 13 places where we're calculate calculating water velocities u and v and and salinity and suspended sediment concentrations in each of those 13 layers in the sediment bed we keep track of the sediment grain size distribution and even though I don't know why John drew in u and v for our sediment bed layers because it's that's a bit misleading it's not going anywhere it's that I'm at bad at the only place it goes is through erosion and deposition it exchanges with the overlying water okay so for each of our sediment bed layers so here sediment bed layer one is the one that's at the sediment water interface and we have in the model we'll use today we have 10 sediment bed layers so n bed equals 10 and so within each of those sediment bed layers the model keeps track of the amount of sediment for each grain size class that's in each of the bed layers if you have erosion of one size class that material would be removed from the surface layer and put up into suspended port in the in the overlying layer deposition of sediment settles out of the water column and goes into sediment bed layer one so it's so so Irene is asking about this variable bed ij so ij is your horizontal location on the grid and then the third index 1 2 up to n bed that's your bed layer and then this other index i thick so this bed variable stores all the information about the sediment bed so the i thick the sediment bed thickness is the thickness of that bed layer and it's tracked in kilograms per square meter and it changes because our surface layer is going to get thicker or thinner depending on the active layer thickness of the bed so that surface layer thickness is going to change and then the way I like to run the model is to have this bottom layer be a really thick layer so that if I erode off a layer at the top I can split off another layer at the bottom okay so yeah the thicknesses of the layers change with time so when the model output will have a bed a model variable called bed thickness that's a 3d variable in time 3d because we have x and y locations we have thickness in the bed and then it changes with time because as the bed layers evolve so it's forward yeah really saying it's 40 because time is a very deep okay now the water column grid this I guess this isn't showing up very well but this is I want to point out that the ROMs vertical grid in the water column is a stretched grid so that's what makes ROMs different from models like Princeton ocean model is that it uses a ROMs uses a stretched grid instead of a Sigma grid so in a model like so in ROMs the the the model is terrain following meaning that the bottom grid cell always follows the bathymetry okay so we don't have a stair stepping grid in the vertical we have a train following grid in the vertical and but we keep the number of vertical grids constant so in the model will do today where we're gonna have 13 water column grids this this figure here that I took from the ROMs wiki has 20 vertical grids and so they're kind of gray lines showing where the grid the water grid cell interfaces are so since you always have 20 grid cells then in shallow water your grid cells are gonna be a lot thinner than in deep water right and so one issue with the Sigma coordinate grid is for a model like Princeton ocean model the percentage of the water column that the grid layer includes is always a constant so a grid layer might be 10% of the water column so in a 10 meter deep layer that grid that grid cells gonna be 1 meter thick in 100 meters water depth that grid cells gonna be 10 meters thick because it's gonna scale with it so that gives kind of a Sigma coordinate model then has a hard time resolving high gradients near the bed when you get into deeper water so what ROMs does differently is it uses a more complicated vertical grid so that it stretches out the the grid cells in the deep water so that it always maintains a thin or you can always maintain a thin layer above the bed right at the bed sediment water interface and a thin layer up at the water surface so in this figure if you can I don't know if you can see the grid lines very well but there's there are thin layers following the bed so that you could resolve some of the concentration gradients and set suspended sediment in deep water because your grid cells aren't getting huge in the deep water then you could also help to resolve some of the surface layer that's influenced by wind forcing because you can maintain thin layers so that's another choice that the modeler when you set your grid up you tell it how to stretch the grid for some applications you might not want to resolve high resolution near the bed you might prefer to prefer to have your resolution all at the water surface usually since I do sediment transport I want to keep resolution at the bed so what are the implications for that is that the that your it's it makes it a little bit more complicated to calculate your depths if you want to plot a profile then you need to get your Z's so it makes it a little complicated to calculate that and on top of the fact that the grid is changing with space the grid changes in time because the sea surface elevation can go up and down so if you have sea surface elevation going up from tides or something then your whole grid has to expand to accommodate that so your bed your delta Z's that are your that are your layer thicknesses change with space but they also change with time and so ROMs doesn't put those Z's those water layer thicknesses into the output file so instead you have to run some kind of post-processing if you want to get those vertical depths and then this is Hernan Arango's sediment transport figure and I thought it was a lot prettier than anything I could have come up with today so I included it it has the transport equation for sediment in the in the top and then again that John Warner figure showing the the sediment bed interface and the water column interface and it points out some of the features of the sediment transport model there's a you tell it how many sediment tracers you want to include so in the model input file you tell it how many sediment size classes or sediment types you want to include this is a little misleading because it says that you can define cohesive and non cohesive sediment tracers the cohesive version of the model is still more under development and so the non cohesive version of the model is the one that's obtained with the main trunk of ROMs for each of those sediment tracers it has a fixed grain size sediment density settling velocity and critical shear stress for erosion so it doesn't account for changes to to either settling velocity or critical shear stress for erosion in this in the non cohesive version of the model but we can trade we transfer sediment between the water column and the seabed through erosion and deposition I said before we have a user to find number of bed layers so that stays constant in the model and the example we'll use will have 10 bed layers and for each of those bed layers the model keeps track of the thickness of the bed layer the sediment size class distribution so what percentage of that bed layers made up from each size class and it says it keeps track of the porosity in the age of the bed layer but those are really just set as constant in the model okay so now this is the case that we're gonna look at and run today this is this case is called River Plume 2 and there's been a version in the main trunk of ROMs there's been a version of River Plume 2 for some time but I used it for teaching the class that I teach at VIMS and and to prepare for this clinic and we had to make some changes to the main the version that comes with the trunk in order for it to work so I don't know if it was bit rot through the decade or what but things changed it and but but with Julia's help we and some of the students in the class also helped to figure out how to make the model run better okay so the way what the model is it's River Plume 2 and it was originally based on a paper that Rich Signel wrote with Jason Hyatt where they they subjected an idealized continental shelf with a River Plume coming out to a bunch of different advection schemes and looked at how well how the River Plume responded to different advection schemes but but this version of the model is pretty different I think than what Rich and Jason worked on back when they did it so the way the model is set up is it's an idealized continental shelf meaning it's just planar the bathymetry is just straight lines that the inshore boundary is 15 on the on the shoreline the water depth is 15 meters on the offshore the water depth is a hundred meters the cross shelf extent is 50 kilometers and we use 52 grid cells to give us those 50 kilometers across the shelf and then the longshore grid dimension is a hundred kilometers what some of the things we added to River Plume 2 was we added waves because the one that comes with the main trunk doesn't have waves but since we do sediment transport of course we wanted to put waves in so we added just a steady uniform wave height of two meters with a wave period of 10 seconds we had and then there's a river discharge coming out 1500 cubic meters per second freshwater discharge with two grams per liter of sediment coming out onto the shelf and then the model is set up to have in a long shelf current of five centimeters per second so at the kind of at the northern boundary this a long shelf current is imposed so that the river so that that helps the that's just and not kind of idealized a long shelf current carrying both the freshwater and the sediment down the coast and then I drew into this little cartoon I drew in a V showing that we get a velocity profile because even we we impose a mean five centimeter per second current but the model solves for 3d velocity fields and then the CS is supposed to represent a suspended sediment concentration that the model is also accounting for the wave direction is just straight in yeah okay this is a little bit more of the details of the model and you know I think that I one I one reason I went into so much details not because I think you'll remember all of this right here but I think this PowerPoint will end up on the systems website and if we if we supply this kind of idealized test case we're going to want this documentation for it so here I outlined what I used for sediment types we included three sediment types and we put two of those in through the river and then the third sediment type the coarser one we just put on this sediment bed and to get the model looking to kind of deal with some issues of the model behavior we made that sediment bed stuff not be eroded so the only sediment that's actively transported in the test case that we have today is the finer grain stuff that settles at point oh six and about six millimeters per second and let's see the I think I've covered a lot of this already so the boundary conditions we had to change the boundary conditions that came with the normal ROMs and I think what's most relevant for for thinking about what's going on in the model today is that the the river flow and the long shelf current are both specified as boundary conditions so they're both specified as kind of point sources of water with the river flow set up to get 1500 cubic meters per second and the along shelf current set up to give five centimeters per second depth average currents and we're using MP data for the tracers with the sediment transporting sediment we found that MP data is needed to have stability in the sediment field the bottom boundary layer model is the wave current interaction of Madsen 94 and in ROMs language it's coded as the SSW option which stands for the the guys at the USGS so what I'm going to just show you what different what you can do with this type of model I have examples from the class that I taught so I have grad students take this class and it's kind of a computer lab class and each each couple of weeks we do a different idealized case and what they do is they first look at the model run that Julia and I got to run and then they make changes to the model run and write up a lab report on what they found okay so I took figures from their work to show you examples of what what kind of stuff you guys can plot with the River Plume 2 test case so here the top panel is the model grid that they used so for this this student used more grid cells than we're using they used a hundred grid cells in the cross shelf and two hundred grid cells in the long shelf so the one we're using is about a quarter of the size and and one reason we went to a smaller size is so to run more quickly and the river mouth comes in at grid cell about about a hundred and eighty here but in the model we'll run it comes in at grid cell 61 and then the boundary conditions that we used in class were really different but I really liked how this student this student by the way got the highest grade in the class and one reason was his lab reports were always so well organized I liked how he provided the boundary conditions in his lab right up so I'm going to show four examples of student work the first two changed the river discharge and then the third played with the advection scheme for the tracers and the fourth changed the boundary conditions of the model so the first example is from Britt Dean she's a biology student at VIMS and she wanted to she had the idea that a lot of our rivers are in drought now and so she wanted to run the model to see what it would be like in drought so she cut the river discharge off and so her to her figure on the top is a time snap from the two model runs with the left panel being the normal discharge of 1500 cubic meters per second and the right hand panel being reduced discharge and so she's plotting with the color the salinity and their arrows in there that represent the velocity and so the main difference is when she you know when she cut the river discharge down her salinity prune got much thinner and also I think the velocities in the plume changed the next model experiment that was done was this student also changed river discharge but they they took the baseline cases their test case number two and they have been kind of doubled the river discharge and then they analyzed how the plume area changed and the how far from shore the plume extended okay and so so here and then they provided two figures the first figure is sea surface elevation because when the plume comes out that influences the density field on the shelf and so you get changes to sea surface elevation so here they're showing the middle is the 1500 cubic meters per second and then the red is areas where the sea surface is elevated relative to mean kind of mean sea level so you see that as the freshwater discharge increase the sea surface elevation in the plume increase okay and then the arrows represent the currents and then the next panel and the same model runs and here they plotted the salinity versus the stream the stream functions and so you see that as the freshwater discharge increases we get a bigger plume but we also start to get a recirculation in the plume that's why so you have those that kind of spiral showing that the the plume is setting up a bus and Eddie and then the third example these students wanted to compare advection schemes and so they had their model run with the left hand panel is using a fourth order advection scheme and the right hand panel is using the MP data and I didn't so I'm a little surprised at how what their results were but I didn't go back and double-check their results so here you see as the as a as a model run goes on it's it's ticking away at time that's day 20 day 22 you see as it as it goes on in time the mob the MP data run gets unstable which to me is opposite of how it's supposed to be but but so what they're plotting in their movie is the salinity field in color and then they're using arrows to plot the velocity and at the top they have their timestamp so now we're up to about day 75 so that's showing suppose so the two models are set up exactly the same the only difference is the advection scheme that was used for salinity and they also set up I like the lots as we look in plan view but it's a 3d model so it's nice to look sometimes at a transect so here they plotted a transect across the river mouth and you know all the action is up where the freshwater plume is in the river mouth and so what you see is that even even we set up the model to have uniform discharge coming in at the river mouth but an estuarine circulation sets up really quickly so that we have kind of the salt water coming in under the freshwater that's going out and that ended up making us trap a lot of our sediment in our little estuary once we put sediment transport in and then the last model run this student Danielle Tarpley Smith changed the the way that the boundary conditions were treated and it and the way that the the way that we're going to run the model is more like her test case where she she changed the the model the way the free surface the sea surface elevation was treated so in the standard model that they ran the eastern boundary was closed and the sea surface elevation was allowed to change but in her test case she clamped the sea surface elevation so that we wouldn't get kind of this large surge of water coming in so here's a time series plot so the top panel shows her sea surface elevation in the standard model you see we got she got about 10 centimeters of sediment added to the not sediment but water added to the water column because of the way the open boundaries were acting but when she clamped that to be to be a small number then the test then that case ran better so the case that we're going to run today we used a clamped sea surface elevation instead of the one that had come in the in the model all right so unless there are questions about that then we're ready to give you guys some of the model output I'm going to show you what some of the model output names are and then we have this file ocean his CS DMS 3 NC that's a model output file and we can put them on your computers either using flash drives or you can download them from an FTP site right it's on the FTP site and then you can analyze the model and after you do that for a little while we can think about what we'd like to run differently in the model and that this is a kind of ideas of of tools that you can use here today we can use MATLAB for looking at the model output if you have access to beach you could run the code today but if you want to try some of this later on your own time if you have access to beach you can run the code there if you want you could take the source code and recompile it to run on your own computer at our in our lab we tend to look to to use some toolboxes SNC tools and NC toolbox to analyze the model people who are fans of Python have developed a toolkit called pyroms and then for more general kind of questions and informations the ROMs community kind of follows the what the wiki and the forum at the ROMs website there on the next two panels I say what's in the output file okay so I have it set up grid information 2d variables and 3d variables so the grid information you've got your horizontal grid points the vertical grid points the water depth and the sediment bed grid and then time is really in time is also kind of great the time grid and so the panel on the columns on the left are kind of wet in English you would call the variable the column in the middle is what the output file calls the variable so it calls the the middle of the grid cells are called x row and y row and then the column on the right tells you the units so x row and y row are stored in meters the vertical grid Z isn't in the history file so you need to run a Matlab script to get those z values the water depth is called h and so the water depth is kind of the bathymetry that's the standard water depth and then overlaying on that is the sea surface elevation that gives you total water depth the sediment bed grid is in bed thickness is the name of the variable and then the time is called ocean time and that's in seconds 2d variables that are stored are depth average velocity so that's u bar and v bar okay then the total bed stress I'm not quite sure who gave it this name but I call it bavuster and bavuster and usually leave off the CW max but but it says it stands for bed u or v component of the stress for the current and wave maximum value so usually what I do is I read in bavuster and bavuster and I square them and add them and take the square root to get something that I call tell b and those are in Pascal's the sea surface elevation is zeta in meters so that's going to be the change to the overall water depth and the wave orbital velocity is called u bot and v bot and then the 3d variables that are in the history file that I mean they're actually a lot more 3d variables I just kind of wrote the ones that I tend to look at the velocities are u v and w and they're going to be at the u v and w points so they're going to not be at the same location they're not going to have the same size so if you want to get total speed which would be u squared plus v squared the square root of it Julia wrote a an M file that interpolates everything to the row point so you can get the total speed the tracer concentrations are salt and temperature and there they are calculated in the middle of the grid cells the sediment concentrations in turn their units are kilograms per cubic meter and they're called mud oh one mud oh two mud oh three which is another example of a I would say a poor choice of name because it makes you think that you're running a cohesive model right but but you're not it's a non cohesive model but the variable names are mud oh one mud oh two and mud oh three the bed sediment you can get the thickness of the layers and is in bed thickness but a lot of times I look at the mud the the fraction the amount of each sediment type in each bed layer is stored in mud mass oh one mud mass oh two and mud mass oh three so these the one two and three are because we have three grain classes if you had seven grain classes you'd have mud mass oh one all the way up to mud mass oh seven okay so if you want to get so that's going to the bed sediment is going to be the amount of each sediment class in terms of kilograms per square meter for each bed layer so if you want to get the total amount of size class one on the bed you can just sum it you sum up all the kilograms per square meter for all of your 10 layers and that gives you the kilograms per square meter of that size class on the bed okay are there questions about that or you might have questions once you get into it okay so now we're gonna give you the output file and you can play around with making some figures maybe for 20 minutes and then think about what kind of different what you would like to change in the model to try and start another model run that can run that we can get running before the coffee break all right so if you want to work in Matlab but you don't have Matlab can you you want to go find a buddy and if you if you have maybe raise your hand if you have to get the I need to go what the next slide says where you can get the data so you could get both the model output and M files that will help you analyze the model output we have it on thumb drives or you can get it try and get it from the FTP site okay so maybe raise your hand if you would like Tara or Julia or me to help you get the data on from the thumb drive Chris I think that might be the quickest easiest approach you guys can you guys I'm gonna get a drink of water and maybe in a few minutes we'll pause this and we can talk about what kind of model runs we might like to do while and if anyone wants directions on how to run it later we're being on the thumb drive files to see what is the cut or well we collect it rolling so we'll figure we're gonna edit you know edit at the dead spots as far as what are you talking about so you're done with her okay I'll keep an eye you're gonna talk at all maybe like you know say like something like that