 Okay, thanks, Stein. Okay, so I'm going to give you a little update on the meeting as it unfolds and systems CSDMS. There may be some of you who are here for the first time. This acronym CSDMS, no vowels, we pronounce it systems. And so sometimes you're going to hear me say systems, I'm meaning CSDMS. Okay, just, okay, here we go. So I've got a few slides that I would just like to go over to, for those of you who are new to the program, to introduce you to what we do and welcome you to our community. So systems offers, we do capacity building and we do community networking. And so you're going to hear a little bit about that. This is one of our goals is to get this community that really didn't have a computational home to work together and work with one another on different projects. So that would be the first thing that CSDMS as a program offers us and others. The second thing is that we offer a model repository. And I'll talk to you a little bit about that. We do middleware, so that's the coupling of these models and model reuse. And we provide high performance computing support, as Stein Stewart just said. And we provide a lot of educational products and knowledge products. And finally, this model support we are offering, and we hope to move forward with this in a bigger way, data repositories, data ingestion into models, other modeling tools and information related to the support of modeling. And so our domain, you see that it's everywhere from the tops of mountains to the bottom of the ocean. There are other programs funded by the National Science Foundation and others around the world that do more atmospheric research and the connections to the ocean and land, such as the community Earth surface modeling community and the cyber infrastructure and geodynamics. We're doing the skin of the Earth, the oceans, the land, the forests and the glaciers. So, these are the latest numbers. You may be interested in that. You're part of a community now of about 650 members. And we have a curve that lets you know whether we're growing or not. I actually thought I'm not the only one. I think those of us that were part of the original forming of CSDMS about 10 years ago, we thought that we would probably asymptote out at around 200 members. We started with about 80 members and we thought we could probably double a little bit more. So we're now at 650 members and if you look at that curve, there might even be an upward tick. So, this has provided us with excitement but also worries because we don't have a budget to support 650 members. We have a budget that supports about 200 members. And, you know, given the country's financial constraints, we don't think we're going to get much more money to support this larger community. So, we need to figure out how we can engage with so many people and yet not offer the kind of financial incentive to show up at meetings and participate. So, please talk amongst yourselves how we do that. But the signs are 650 is where we are today, but this is likely going to hit 1,000 soon. And in addition, we had working groups that only had, you know, 50, 60 people at most in them and that's a manageable number. Now, we have working groups that themselves are huge communities on their own. And one of the things that we have to make sure we don't do is go back into our silos. CSDMS is about the moving beyond the boundaries of people who work just in the terrestrial sphere to reach out to talk to those in the coastal and marine communities. So, even though these working groups are now getting very large, 300 plus in terrestrial itself, we have to worry about that. So, CSDMS partners with its members which come from over 250 institutions, 125 in the U.S. These numbers are hard to keep up with, so these probably old numbers. But there are about 95 universities in the U.S. About 12 private corporations in 19 government labs that are participating. And in the world, there's probably members from 40. Marlene said 41 the other day. Countries, 85 universities, three private companies, and 27 government agencies. So, you've got government, you've got industry, and you've got academics working together to try to deal with the issues and why this all formed in the first place, such as people keeping models to themselves, never sharing them, trying to possibly not let people know the algorithms that are behind how they did their simulations. We're trying to avoid all of that. So, in the last four years, four plus years, we've sponsored or hosted 120 workshops, symposium meetings. We put on 11 short courses in a number of countries. Our staff has gone around to fly the flag and make presentations. And we've written peer-reviewed journal papers. But this is just the small staff that has accomplished this. And just to let you know, we did a track on how many abstracts are going to be given as presentations and posters at AGU coming up in December, and it was over 200 in that one meeting. So, actually, the community itself is writing many, many papers and presentations that we could all be part of. So, it's a large community, a lot of information being passed around to one another. So, if you want to know the size of our staff and who's paying for it, we have 5.6 full-time equivalent positions, and some of these, most of these are only part-time. And we fund the rest of their positions through other sources. And then our third supply is other funds that are funding our graduate students, since there's no money available through the NSF Initial Cooperative Agreement. We have a lot of graduate students coming through through other support. We have only a couple of undergraduate students. We hope to grow that in the future. We funded three postdocs, and there are two senior research scientists that are also through other funds that are at the facility. And one of the things that we do, and it takes a lot of time of our staff, is to have a lot of people coming and visiting our facilities. Many of them are from the U.S. per se, but we've also had a visiting scientist from Canada, Germany, Norway, Austria, Italy, Colombia, China, the Netherlands, and Belgium. And these come, they also, they do take some of our time, but they also contribute to the effort and to help get their models up and running and available to all of you. So, where are we with our model repository? So, the numbers keep growing, and I think that's a good thing. Again, with growth comes extra work, and maybe I should tell you a little bit about that. So, we have about 140 numerical models that you can download, and don't forget we're all open source freely available. So, some sites, we send some of the people the downloads to other sites because they would like to track how many people are using their models and register with their own software community. So, for instance, if you go to our site, you click on ROMs, we will send you to the ROM site. So, we offer 140 models through our own repository that people have provided us, but there are other models that you can still get through our site, but you will be downloading from another site and registering through another site. We provide a lot of tools that support these models. They're not quite simulation models. They're models or tools, numerical tools that will help those models go to the next level, and you can see how they're distributed there. And we're trying to get these models slowly but surely into what's called plug-and-play so that they can be coupled with one another. That's the middle where work we do. And we have about 53 components now. So, if you go to our site and you start drilling into these models, you will get a lot of information. And it's that secondary information that hopefully will help you. Getting the code is only one thing, but learning what the code does, the references, the equations, and all of that are what makes this code come alive. In terms of downloads, a little graph there shows you the kind of downloads we get. Obviously, wintertime, Christmas time. People seem to go into their computers at home, tired of the holidays and download models. So we have, not counting this year, last year we've had up to 7,000 downloads. And so I don't know where we'll be at the end of this year, but we have a little table here for you to look at some models. We have about 14 of them get a lot of downloads. They're obviously popular. And this provides us with feedback on which ones that we should put effort into making and to plug-and-play. Often these top 14 models, they're large. They tend to be on the large size. They're complex. And they probably take longer than some of the other models to componentize. So it's you who are donating your models can think of this almost like a citation index. You're getting confirmation of who's interested in your model. So if you're one of the less than 20, if you go on to the site and you say, oh, not many people are downloading our model, there are many reasons why that might be. Number one, you may have just given it to us, so it will take time for that to get penetration into the systems community. Number two, you may not have published any papers or not many papers or your community may not have published many papers on this model and therefore it's not really well known yet. So hopefully by the time we get to it and we componentize it, they will gain some momentum and other people can start using it and hopefully that should allow you to take some pride in the effort you put into this. I feel very strongly that in the university world, they value, this is for my own vice chancellor, they value the number of papers that you publish that are peer-reviewed. So it's really important for university administrators, government, lab supervisors to understand the effort that goes into putting out a model and even though it may not be considered peer-reviewed in the normal sense, through our site every bit is valuable as a peer-reviewed paper, probably taking a lot more effort and should count equally if not more than published paper. So let me go on to the types of models that systems is now having under its umbrella. So the first type are what we call landscape, seascape evolution models. So these are the stratigraphic geomorphic models that simulate across geological time and space. So we're not talking hundreds of years, we're talking thousands if not millions of years. And they incorporate a variety of boundary physics such as geophysics and geochemistry feedbacks including sea level, climate change, tectonics, post-depositional processes and that sort of thing. So there's a large community within CSCMS that does this kind of modeling. And you're going to hear some talks on this today, tomorrow and the next day. The next type is morphodynamics. So this is where the flow interacts with the bed. The bed may form bed forms which changes the characteristic of the flow and this dynamic feedback between the flow and the bed is what we call morphodynamics. It's extremely popular in the worlds of engineering and sedimentology and you're going to be hearing a number of great talks on that. You know, we now have a seat on the River Coastal Estrine Morphodynamic Organization run by the International Association of Hydraulic Research, IAHR. And we had a meeting in China this most recent year. There's one and another two years in Spain. And if you go to these meetings, you will see that, you know, this community is moving strongly into echo-hydrodynamics. So they're reaching beyond their comfort zone to include ecology and I think this is a thrust that we should all pay attention to. Then there's the transport and circulation models. This is a large community. Typically you may think of this as the ocean circulation people but it also could be the hydrology people that are moving things from place A to place B. They are not really paying attention to what's happening to the bed. So the ocean changes on the bed whether bed forms develop or not are not really impacting the ocean circulation itself. And for the types of models that are done in hydrology where you may be moving nutrients from place A to place B, you're not again interested or you're not always modeling the details of what's going on in the bed. You're just moving it from place A to place B. So this may be perhaps the largest number of models that are out there. They're very useful to the kind of research that goes on in earth system science and you'll hear a little bit about that too. So here's the various domains and the kind of models that we have. You don't need to read the details. You can go to our website but I've broken it out into the various environmental domains. This is terrestrial, hydrology, coastal marine, and then miscellaneous models where you have climate, weather and tools and things like that. But if you drill down below those model categories, you will see that each of these could have in the terrestrial world, you could have landscape evolution models, you could have transport models, and you could have morphodynamic models. So this is the complexity in the landscape that we have, we can work with. We have, if you don't include some of the unnecessary lines in code, we're looking at four to five million lines of code. So I don't know how many Leo Toadstoy novels that would be, but I would imagine it's like 10,000 or something like that. So this is the level of code that I don't expect any of you to be reading for your going to bed at night. But it is open source. You can get drilled down, find out how other people have done their things. Maybe you think you've got a better way of doing it. Maybe you want to change the way those folks are coding or have coded their algorithms up. And that's what it's all about. You know, go into Subversion, get there, see what's going on, maybe make some changes, tell the people who originally came up with the code what you're working on, that's the community effort. So as Stein says that we've, we, the University of Colorado has firmly grasped what's going on in CSDMS and are supporting all of you. And we're supporting you through a couple of supercomputers. So this one is funded by the University of Colorado and it's funded by the U.S. Geological Survey. It's a eight-teriflop supercomputer. It's called Beech. It right now has 150 of you folks as members working on running models. And for you to do that, you have to meet one of the use criteria outlined here. You're going to advance science with one of the systems models. You're going to develop a model, not yet on the repository or you might be involved with data systems or data visualization. So it's free. And I guess there's not many things in the world today that are free. And what comes with that often is support for you and your modeling effort through the integration facility. On behalf of the National Science Foundation, our community has been identified as a community that's right to move to the next level. And that is large-scale supercomputing. And so the National Science Foundation has, along with a lot of money from the University of Colorado, has built this computer called Janus. This front-range computer was in the top 25 supercomputers in the world eight months ago. I don't know what it is today. It falls every day after you launch it. But it's got 16,416 cores. It's a big machine. More than 150 teraflops. It's starting to be extremely well used. So you won't be able to get all of these resources to yourself. But if you're interested in running some models, we have a number of CSDMS members running models on it. We can make that available for you just work through the integration facility. And along with this computer, one of the clinics that we offer at this meeting is some, and we will continue to do this, is to help our community who really is not used to doing high-performance computing to learn to code in that manner. So we have this data repository. We focused on initialization databases. These are mostly global databases that you can get. We've got 50 or 60 of these that you can get access to the data to initialize your models or run your models. In the next five years, we want to move beyond that. And if we have a chance to talk amongst yourselves, then I would focus on these test and validation data and benchmarking data. I think we really want to do this as a community. If you have a model, how good is it? And even if you don't know how good it is, then how does it perform against another model? In other words, neither model. There's no real way of knowing how well they do, but at least it'd be good to know how well it does against another model. So one's called benchmarking and one's called validation or testing. So both of these, I think, are ripe fields for us to move into and to work with the people who are doing experimental work in labs and with their data, which is probably the best way to test models, but also the field data that are well characterized that we can use to test these models. And you're going to hear some of that in some of the talks today. We have the Kowasi hydrologic community represented here. And then we have this education repository where we've got some fame related to it, as you'll see in a moment, but we offer a lot of simulation movies, real event movies, lots of student labs and short courses, lectures, textbooks, imagery that are now being consumed by professors and their instructions for their instruction of students and universities around the world. So there's two ways to get access to it. You can get access to it through our website. That was what these slides were showing you in the images directly through our website. We have a system that has a YouTube channel. And you can go to the YouTube channel and be one of the people who downloads it. We are in the top 50 every once in a while for the number of people who are reaching into a nonprofit category and getting access to the simulation and real event movies. So if you have... This is trying to reach out to you. If you've got a movie and you don't see it on this website, please donate it. We will ascribe where it comes from and give you all the kudos for it, but you certainly can get penetration. And if you want to do well in your outreach, at least as tracked by the National Science Foundation and other U.S. agencies, this is a good way of doing it. Get your information on this site. We have a number of instructional videos that will handhold you. You can go to the site, either the YouTube site or the website, and they'll walk you step by step on how to... Let's see how to connect to our supercomputer, how to contribute to the system's repository, how to use the model repository, how to become a member. And sometimes it's a little embarrassing to say, I don't know how to turn my computer on, especially if you're supposed to be one of these top-notch supercomputer people, but there you go, and you can do it in the privacy of your home. So you laugh. Okay. So, some of you have seen this over the last few years. We work on tweak these protocols all the time. We work with the Computers and Geoscience, which is one of the journals of the International Association of Mathematical Geoscientists, on these protocols because they think our protocols should be their protocols, and we keep working them. So you'll notice language changes from time to time as we reflect on what we really expect. I think maybe possibly at the beginning we were too onerous. But right now we would like the contributed models, of course, to be open source, but we'd like them to be licensed. So if you go to our site, we will help you license your models, and if you don't license your models in an open source, we will license it for you. But we would communicate with you so that there are many open source licenses, but once your model is licensed, it helps everyone who wants to use it know how and when and if they can use it. We definitely want it to be widely available. So you will have seen, we've got rid of about 40 models on our website that used to consider themselves open source, but they wouldn't be freely available. So if they're not freely available, it doesn't matter whether they're open source. If people can't get at it, or you are controlling who gets the models. So through our site, or we make sure of these community sites like the ROM site, that they really buy into this open source, and therefore the downloads, there could be somebody who wants it, but because they happen to be of a certain height or whatever, they can't get a hold of the model. So vetted. So we still struggle with this one. We'd like the models to be vetted. I mean, what's the point of providing us with a model if it's never been vetted, never been tested? No one's ever run the model outside of the person who's used it. So our community is supposed to do this, and I don't think we've really put too much effort as a community into vetting these models, and I think that's something we need to work on in the future. So this doesn't have to be an onerous task, but if you say my model has water run downhill and someone runs it and it doesn't run downhill, you know, it doesn't meet that criteria. So that's the level of vetting. It should do what it says it does. But there are more advanced levels of vetting, you know, running against benchmark or verification test sets and things like that. That's sort of another level that hopefully someday we can get to. We'd like it just... If you wanted to... You're the models to be in one of our plug-and-play components. It has to be in a Babel-supported open-source language. So that's C, C++, Fortran, Java, Python, or have a pathway to conversion. So if you go to our site, you will see models that are written in languages that are not one of those. So that's fine, too. So we have MATLAB models. We have some Viz models and SAS models that are on the website. But we can't use them as a component or componentize them simply because they're not Babel-supported. So that's all that that means. We certainly will accept the models. We'd like the models refracted with an IRF interface, and I'll tell you a little bit about that. And we help people do that. We'd like the GUIs that are already associated with the model separated out because the GUI that comes with the model likely we won't use it. So when someone downloads the model, they could download the model with the originally developed GUI. That's fine. However, in one run through our system, all the GUIs look exactly the same. That really ups the level and penetration of the model into the community. We'd like it to come with a well-described metadata file. A lot of the models that first came in didn't, and we had to fill them out. And so we go back to the original author and say, is this what this model really does? Is this the equations that really used, et cetera? And we'd like the model to come with test input-output files so that when we compile it and we run it and others compile it and run it, it will do what the author thinks it does on another system. We'd like the model to be clean and documented and all dusted off looking good. And this is kind of important. We'd like the state variables to be well-described with identified units. If we don't know what the units are, it would be hard to have them in a plug-and-play world. So I see Chris Sherwood in the audience. He helped me and others in the room help me put why open source is so important. And this is, I'm showing this not just because it's sort of the fundamentals behind our approach and our community approach, but I think if people ask you why should your models or their models be open source, I would like you to be able to explain it to them in this coherent way. So revealing source code provides the scientific hypothesis embodied in an American model and reveals their implementation. That's important. That's what science is all about. Details are important. A solution to a set of equations can take numerous forms. Each solution has its own pyramid of assumptions and limitations, and so you want people to understand how you got to where you got. And so this co-transparency is the foundation of modern science, we think, and allows the replication of science and results. And so if you, just like peer-reviewed papers, have gone through an openness, where is the data, how did you analyze your data, all that stuff, the same should be done for our modeling community. Anyways, moving on. So here's how it works. We get a model, we'll call it an initial standalone model. The very first thing that we do at the integration facility is that we will compile it. If it doesn't compile, we don't really do much after that. We may go back to the author and says, you know, it doesn't compile, but we try to, that's the very first thing we do. And once it compiles, we will take the input-output files that came with it and we will run it and see if we get the same results. When we do those two simple things, we will then examine whether the metadata file is complete and if not help complete it. And then we will see if the references are there and all of that. We then release it to the community. So that's just a typical download standalone model. It will have gone through all those tests. The community can use this model on our website because it compiled on our website. And it's available through our website. Okay, so then we have this model. How do we make it into a component? We will make sure that if it hasn't come in with an IRF interface, we will give it one or work with the author to give it one. And we're getting very good now at describing this in terms of what a person needs to do to give it an IRF interface. We call it a basic modeling interface, a BMI. So simply breaking the model into initialized run finalize. Many models already come that way, but we make sure it is in that form. To take it to the next level, can the model couple with another model? That's a different world. You will be able to run your model within the CMT tool, but it may not be able to couple with another model. To do that, you need to be able to tell that model what is in your model, what models it can set, another model, and what information it can get. And so you can reach into another model, maybe change some stuff, reach back into the model, and maybe provide some information. And this reaching in and out of models is different than simply input-output files. If it was just input-output files, this would have been done years ago. So this is a little bit more complicated, and we call this as our CMI, our system modeling interface, and it's much, much more complicated. But that's how we get these models to be plug-and-play. When it's beside another model, that has this CMI interface, then that doesn't mean these models can work together, because maybe they're not designed to work together. So not all models with a plug-and-play will hook up. Only ones that should hook up can hook up. So these were the design criteria for our framework. We support multiple operating systems. We want it to offer language interoperability, deal with structured and unstructured grids, our GUI to be platform-independent, the graphic packages to be platform-independent, to meet open-source standards, to be able to work in a parallel computational world, be friendly with other efforts at Middleware, and to provide familiarity with how things are done. So most of you have seen this diagram. I'm not going to go into it in any detail, but we're all about this plug-and-play programming. And this is for the first few years of CSDMS. This is where a lot of effort has gone in to come up with this system. Now that the system's in place, it's just work getting the code that are contributed into the library. So this component modeling tool behind it is just huge codes. And we have some of the people who have written these codes in the audience today. So from Department of Energy, we are using the suite called the Common Component Architecture. Within that is Babel, the language interoperability compiler. There's Boca that deals with component preparation and program management. There's caffeine that deals with low-level model coupling within high-performance computing environment. We use various re-gridding packages. We use the OpenMI standard re-grid and the multi-processor re-gridding package, ESMF. We are aware and try to incorporate the OpenMI standards, which are pretty good interface standards. We don't follow it exactly, but we pay attention and work with that community. And we have Mike Ellis in the audience who is now putting the British Geological Survey effort behind OpenMI. We work with data formats. NetCDF is the standard that we like and try to make sure that we are all the models, pump that out. We also work with the quasi-world in their water mock-up language. And I think you'll hear about that from John. And our visualization packages uses the high-performance computing package visit, and we have a clinic on that. And we are working in the high-performance computing world using MPI, PETSY, OpenMP, and we have a clinic on that. So this is as simple as it gets. Plug and drive. And if these models have a pathway, they can be coupled. And you will see that you will get... There's a remote working directory that tells you where you are in the high-performance computing world. You will put in from the pallet a driver. Almost all the components can either be a driver or just a plug-and-play model. They can be either one, but you will put one of them there, and that's the one that will drive the rest of the components. And you drag it from a pallet. This arena is how you will do this coupling. And if you have models that can talk to one another, these colors will light up. So if you put in the CEM, which is a coastal evolution model, and it needed a river model, you found a river model from the pallet and dragged it there. And even though whatever was labeled discharge, it doesn't matter. If it provides the river input to CEM, it will light up and let you know it's working. And the waves the same thing. So we're trying to get this very approach about, even though the code is complex, to get this very approachable for students to get in this and use this in their own research and education. But we want to avoid black box syndrome. So behind all this is this config button that allows all the input, output files, and all that stuff to be there. It gives you constraints on limits on the values that you should use. And if you go down to in the config button, you'll see a help button. That help button will then open up a website that has all the equations listed out, the references and description of the model. Because we really want people not to fall into the black box syndrome trap. I'm going to finish up shortly, but I want to say a few words pressed by my colleagues at the integration facility. We'd like you to vote, go to our website, and vote on which, provide us feedback through this voting mechanism on which models you would like us to componentize first. And when we finish one up, we will go to this and say, well, this one seems to have the most need. People have voted on it, so we're going to put effort behind that. Otherwise, without this kind of thing, we make our own decisions, and possibly we're saying we should be componentizing a model that no one's interested in us working on. Also to be transparent, just so that everyone sees what we do, you can go to the website and you can see what we're working on and what stages things are. Here's an example. This is a flexure model. Eric Hutton is the project owner, and there are a number of tasks. The tasks maybe provide metadata. Who should own that or is working on that task? More information on that task. Estimated or completion date. And when that's all done, you'll see a check mark and we go down until we actually get a component. That way, people can see where we are in the world of our activities. So I think it's my job on behalf of the community to try to raise money for computational science and engineering. And so this is to the world that we live in, CSDMS. We do a lot, but we do a lot with a limited budget. And just to let you know what our budget is compared to maybe some other things that you know and admire, the National Earth Surface Encent. Yeah, I used to know what that is. National Center for Earth Surface Dynamics, Encent. That's their budget for a five-year period. And the Community Climate System Model, or now as it's called the Community Earth System Model, that's their budget for a five-year. So don't expect us to complete things as these other large software and experimental lab facilities are able to do given the size of our budget. We, of course, try to grow that budget through funding sources, through industry, and other federal agencies. And hopefully over time, even within the National Science Foundation. And for... This is a reminder to the U.S. folks in the audience. We're not doing so well in science and math education. We rank 51st in the World Economic Forum. I think that in general, we need a lot more money in the kind of work we do if we're going to try to get to an end-to-end solution world as we deal with the problems that we have before us over the next 100 years. 6% of the U.S. degrees are in engineering compared to 20% in Japan and 60% in Germany. So I think in the U.S. we can do better. So... So where are we going into the future? We want to work on expanding our reach and multiple platforms, et cetera. We think it'd be really good for all of the models to work on other computers other than our own. And in fact, if ours goes down like it did last year, they could take over, at least for a short period of time so that there will be no downtime. We think we need to move strongly into the coupling physical, biological, and human processes. And so a discussion amongst all of you for this meeting is should we grow in our focused research groups? Possibly having one in geodynamics and other in ecosystems, maybe one in biogeochemistry, you know, and others. That means our community grows even though our budget doesn't grow and there's no impact. That came from the San Antonio meeting. And so did this one from last year. We believe we need a landscape and to rock initiative to do and support the deep time space research that is needed by our energy partners. We also need to move fully into the global environmental change and have our models be more appropriate for sustainability science. And within that, systems is a strong supporter of the concept of international year of deltas. We have a working group that's focusing on that. And if you read the last EOS newsletter, you will have seen an article that we helped put together. We believe that we need to move into the modeling for operational needs. We believe right now we do research-grade code, but eventually this research-grade code we'd like to hand it off to others that would use it in a more ongoing way. So we have a project, Ballash is here, with others taking NASA products and coupling them to system models. We have another one that's going on with the Bureau of Ocean Energy Management, coupling system models, and of course our support of the Chesapeake Community Modeling Program. Again, we want to do more with benchmarking and model-intercomparison. Model-intercomparison hopefully will be a thrust of the new systems next five years. And of course we want to support and continue to support field work and programs. And in this specifically, we've signed agreements to support the work that's going on by the NSF-FESD Delta project. So take us to this final slide. For the rest of the day, we're going to have keynote addresses and concepts in the morning. We're going to have an hour lunch. We have to hold to that time. We're going to deal with clinics in the afternoon. There'll be a short break in the middle of those clinics. Then we're going to have poster sessions. And we'd like you to vote on the best poster. We'll announce that at the banquet. On the third day, we're going to have the working groups meet. And on the fourth day, this only concerns some of you, we will have our executive and steering committee meetings. I would like to take this opportunity to thank the University of Colorado and the National Science Foundation. And if you're enjoying the refreshments during the posters and the lunch and breakfast, that is thanks to the Bureau of Ocean Energy Management and Shell. And Shell is also sponsoring the banquet tonight. Thank you very much.