1 2 3 4 5 SAVANNAH HARBOR IMPROVEMENT PROJECT 6 7 STAKEHOLDERS EVALUATION GROUP (SEG) MEETING 8 9 JUNE 3, 2003 10 9:00 A.M. 11 MIGHTY 8TH AIR FORCE HERITAGE MUSEUM 12 SAVANNAH, GEORGIA 13 14 15 16 17 18 19 20 21 22 23 24 25 2 1 2 3 4 5 I N D E X 6 7 OPENING REMARKS AND INTRODUCTIONS ------------- 3 8 FACILITATOR'S OPENING EXERCISE ---------------- 5 9 10 SCIENTIFIC BRIEFING --------------------------- 7 11 BRIEFING QUESTIONS ---------------------------- 40 12 GPA UPDATE ------------------------------------ 55 13 STATUS COMMENTARY ----------------------------- 58 14 MITIGATION ------------------------------------ 73 15 AQUIFER COMMITTEE FUTURE ---------------------- 74 16 PEER REVIEW ----------------------------------- 76 17 REPORT UPDATES -------------------------------- 77 18 NEW BUSINESS ---------------------------------- 81 19 CERTIFICATE ----------------------------------- 86 20 21 22 23 24 25 3 1 INTRODUCTIONS 2 MR. DYSART: Okay. We'll call the meeting to 3 order, and the first thing I'd like to do is have 4 everyone introduce themselves, say who they're here 5 representing, and we'd like to start with 6 Mr. Griffin, go around that way. 7 MR. GRIFFIN: My name is David Griffin. I'm 8 with GDOT. 9 MR. PHILLIPS: John Phillips, Georgia DOT. 10 MR. SUTLIVE: Charlie Sutlive, Savannah 11 Maritime. 12 MR. BEASON: Fred Beason, Bottom Line Echo. 13 MR. BROWNE: Tommy Browne, Savannah Pilots. 14 MR. EUDALY: Ed Eudaly, Fish and Wildlife 15 Service. 16 MR. PLACHY: Doug Plachy, Corps of Engineers, 17 Savannah District. 18 MR. FLOCK: Allan Flock, Savannah National 19 Wildlife Refuge. 20 MR. ROBINETTE: John Robinette, Savannah 21 Coastal Refuge. 22 MS. CLARK: Sabrina Clark, U.S. Fish and 23 Wildlife Service. 24 MR. BROWNING: Donny Browning, U.S. Fish and 25 Wildlife Service. 4 1 INTRODUCTIONS 2 MR. BERSON: Will Berson, the Georgia 3 Conservancy. 4 MR. KEEGAN: Larry Keegan, Lockwood Greene 5 Engineers. 6 MS. SCHALLER: David Schaller, Georgia Ports 7 Authority. 8 MR. REES: Morgan Rees, consultant for Georgia 9 Ports. 10 MR. KYLER: David Kyler, Center for a 11 Sustainable Coast. 12 MR. SCANLON: Bob Scanlon, City of Savannah, 13 and also the Savannah Harbor Committee. 14 MS. HODGE: Jeanne Hodge, the Corps of 15 Engineers. 16 MR. PARKER: Jim Parker, Corps of Engineers. 17 MR. BAILEY: Bill Bailey, Corps of Engineers. 18 MR. DARBY: Jim Darby, the Sierra Club. 19 MR. DYSART: Ben Dysart, SEG facilitator. 20 MR. PARSONS: Keith Parson, the Georgia 21 Department of Natural Resources. 22 MS. LEFFEK: Teri Leffek, Stevens Shipping. 23 MR. STAFFORD: John Stafford, Ogeechee Audobon 24 Society. 25 MR. WRIGHT: Tom Wright, Palmer Johnson 5 1 FACILITATOR'S COMMENTS 2 Savannah. 3 MS. RUTHERFORD: Freda Rutherford, Tybee 4 Island Beach Task Force. 5 MR. OFF: Lou Off, Tybee Island. 6 MR. CONRADS: Paul Conrads, USGS. 7 MR. HALL: Carl Hall, citizen, retired DNR 8 regional supervisor. 9 MR. DYSART: Okay. Thank you. I'd like to 10 welcome you. I got here about 8:30. Our guest 11 speaker was here wanting to know where everybody 12 was. I said, people don't get here till right at 13 9:00 o'clock, and they pour in. 14 It seems like there's more people here on time 15 today than usual. That must be attribute to either 16 the fact that we have a good outside speaker or the 17 fact that Will Berson woke up with a song in his 18 heart, reportedly, in the outer area here. 19 So we've got a lot to look forward to here. 20 You have before you now the proposed agenda for the 21 meeting today. It has items that have been 22 requested by members, some new things, some things 23 that have been requested that they be standing 24 items, and I would ask are there any additions, any 25 changes in order, or anything that you would 6 1 2 desire? 3 If not, I presume this is acceptable and we 4 will operate off of this. I do continue to 5 appreciate the advice and input from members of the 6 Stakeholders Evaluation Group helping me put this 7 together each time. We will consider that 8 approved. 9 The April transcript has been posted. You've 10 had an opportunity to review that. Any comments 11 about that? Doug Plachy. 12 MR. PLACHY: There's a couple of locations in 13 the transcript where last month or two months ago I 14 was talking about the waterways experimentation 15 Station, which the acronym is WES, W-E-S. I 16 believe in the transcript it's listed as W-E-S-T as 17 in the direction west, but I'd forgot to bring 18 those pages with me where that is. 19 MR. DYSART: Okay. Any other comments for the 20 record? In that case, with those changes and that 21 clarification, we'll consider the April meeting 22 record to be accepted. Thank you. Would you 23 introduce yourself? 24 MR. BECKMAN: Yes, sir. I'm Lee Beckman. I'm 25 the new manager of legislative affairs for Georgia 7 1 SCIENTIFIC BRIEFING 2 Ports Authority. 3 MR. DYSART: Thank you. Welcome. As is our 4 custom, anytime we have an outside scientific 5 briefing or studies, what not, we give the speaker 6 the opportunity to be heard promptly, as opposed to 7 waiting around till who knows when. 8 Of course, you're welcome to stay and listen. 9 However, Will Berson had requested that we have a 10 briefing on data mining. So we have today Paul 11 Conrads with us. 12 Would anyone like to make any comments, 13 introductory comments, about why this was requested 14 and so forth -- Will or anyone else -- Larry? 15 MR. BERSON: It's a method being employed. 16 It's one I don't particularly understand. 17 MR. DYSART: Okay. I'm going to say there's 18 been a lot of interest in this topic, and it was 19 requested that someone who was extremely 20 knowledgeable be brought before the group to share 21 their wisdom with us. 22 We're very fortunate to have Paul Conrads, 23 USGS in Columbia, with us today. The floor is 24 yours, sir. 25 MR. CONRADS: Well, thank you. I was putting 8 1 SCIENTIFIC BRIEFING 2 this talk together several weeks ago, at least in 3 my mind I was putting it together, and decided I've 4 been working with neural networks and data mining 5 for about six or seven years. 6 When I make presentations on it, it is 7 fascinating what results you can get with it, but I 8 guess the best analogy I heard somebody saying was 9 this is the like drinking from the fire hose. It's 10 just so much information and kind of lofty, I 11 decided to kind of bring it down to a simpler 12 level, kind of go through the history of what 13 exactly this data mining is all about, show some 14 examples on some work up on the Congaree Swamp, and 15 then on Savannah, and open it up to discussion. 16 It's a very informal thing. Stop me, ask me 17 questions. Anybody can ask me questions to stop me 18 except for the stenographer. 19 What I want to do is talk about data mining 20 itself, and then go in about neural networks, a 21 little bit on the history of neural networks. It's 22 really just come to forefront in the last five to 23 10 years, but actually it has quite a long history 24 that goes back to the 1800s with these things, the 25 description of the actual kind of fundamental 9 1 SCIENTIFIC BRIEFING 2 network, and then go in and talk about some of the 3 applications, and finally talk about the model to 4 marsh study that we're doing. 5 This first quote, I think, kind of sums up, is 6 kind of a good definition of what data mining is. 7 A search for valuable knowledge in massive volumes 8 of data 9 I think when you look at the computing field 10 of the last couple decades, especially probably the 11 last 10 years in particular, these large databases 12 have become prevalent everywhere. This isn't just 13 environmental sciences, it's all the other sciences 14 in other fields. 15 This huge amounts of data, and it's a task of 16 how do we get the information we need out of these 17 databases. 18 One issue that I hear a lot is and maybe this 19 is from a modelling background, data collection 20 background, we can bring truckloads, handtrucks 21 full of data. People say, where is the 22 information? There's all this data and no 23 information. 24 Well, this is a way of going about it. I 25 guess being with the GS, we collect a lot of data. 10 1 SCIENTIFIC BRIEFING 2 Being a modeler, I'm a poster child of lots of data 3 and no information. I'm trying to change my ways 4 here. 5 Data mining itself is a very broad field. 6 It's really a tool box of processes from the 7 various fields including signal processing, 8 advanced statistics machine learning, artificial 9 neural networks, chaos theories, advanced 10 visualization, and it could go on. 11 But it's really taking these techniques of 12 these various fields to try to extract this 13 information. So neural networks are really just 14 kind of one tool in that tool box. Ordinarily, 15 these lead squares and regression, those sort of 16 things can be in there also. 17 There's a lot of different things that can be 18 in there, but it's kind of combining these 19 different resources there. 20 How is it done? Well, what you are really 21 doing is trying to learn and quantify important 22 cause and effect relationships in your data, so 23 that somehow, in the environmental field at least, 24 that's what I'm familiar with, is that the physics 25 of the system is manifest in the data itself. 11 1 SCIENTIFIC BRIEFING 2 So the idea is how do you extract that 3 information from data to reveal what the physics 4 are. 5 Create computer models that are developed 6 directly from this data, so that once you 7 understand that cause and effect relationship, you 8 can build models and simulators that actually 9 become almost like virtual processes of the system 10 itself, so that you can take the variables you have 11 some sort of control over, and actually be able to 12 play the what if scenarios over predicting future 13 outcomes, evaluate alternatives, show the best path 14 forward. That's what we're all interested in. 15 There's a -- I kind of broke this into basic 16 data. Data has properties that must be measured 17 for optimum use. I think this is where the 18 strength comes. We have, you know, uni-variant, 19 you know, one variable, multi-variant relationships 20 interacting with each other. 21 These signals can usually be broken up into 22 kind three different components within the same 23 signal. A tidal signal is a great one to imagine 24 here. 25 You have the periodicity of the orbital 12 1 SCIENTIFIC BRIEFING 2 mechanics going on in the tidal signal, but you 3 also have some chaos on, especially us involved in 4 the government work, we relate to chaos quite well. 5 The best way to think about chaos, without 6 getting real personal, is it's really what can be 7 predicted on a very short horizon. The weather is 8 a great chaotic system in that we can make fairly 9 decent predictions of what's going on in the next 10 day, but we all look at the five day forecast. 11 We all know day four and five, really, is 12 anybody guess, and then there's just basic noise 13 which really is of little use or unpredictable to 14 great degree. 15 So it's really being able to pull out these 16 components of data through there and understand the 17 dynamics. So just think of the complexities of 18 coastal systems, things are happening on incredibly 19 different time frequencies between just hourly 20 tidal changes through seasonal changes. 21 And then there's dimensions of space and time, 22 you know, how things are moving over the years and 23 locally and to a far range also. 24 So that's just information that just can be 25 extracted from the data itself. That mining is 13 1 SCIENTIFIC BRIEFING 2 really how do you pull that information out. 3 Hopefully, by the end of talk, you'll have a 4 little better understanding of that. Also what 5 happens here, it automates discovery. This is one 6 of the big powers I have found. Usually, you go 7 into a new system with some sort conceptual model 8 of how things work. 9 What I've learned, after doing this for about 10 six or seven years now, is how little I know. You 11 go in there and when you actually start saying -- 12 have the data revealed on how the system works, how 13 the physics work, it completely changes your 14 preconceived notions of how the system works, 15 We did a study on the Beaufort River, you 16 know, put out gauge stations. You kind of know 17 which way the flows are going. It's upside down. 18 It turns out the flow is going north, the opposite 19 way, but models that were done 15 years ago all 20 have the water force the other way through the 21 system. 22 So you can really just let the system itself 23 reveal itself to you. Here you can really 24 integrate your data with your professional 25 activities that these models can be set up with so 14 1 SCIENTIFIC BRIEFING 2 much information coming in now, information coming 3 in real time, these models can be set up in real 4 time you can actually have the model sit there and 5 run as soon as the new data is brought in. 6 I think a real future here is with integrating 7 datas is coming in. It really does -- it tells you 8 what it is possible to know. A good data mining 9 job really reveals what information content is in 10 the data set. 11 If it's not there, you still think there's 12 something there, you probably just don't have 13 the data to show it. 14 Again, a lot of behaviors manifested in data, 15 there's a periodic kind of linear system going on. 16 Chaotic, we'll talk more about this. I have a 17 great example if Savannah was a chaotic system 18 going on, and the random, which is mentioned 19 earlier, is fairly unpredictable. 20 Artificial neural networks is just one tool in 21 the data miner's tool box. I think the reason it 22 gets utilized so much is it an allows for 23 non-linear simulation, where lead squares, 24 regression is fairly linear. Most navigable 25 systems manifest some sort of non-linear behavior. 15 1 SCIENTIFIC BRIEFING 2 It will pick it up and is quite powerful. 3 Now I'll be showing examples of that just a 4 little bit. Here's the early theory, you know, 5 going back kind of digging up some of this work. 6 It's kind of interesting. Back, I guess -- again 7 this will be quite brief -- back in the 1870s, in 8 that time, that's when they started looking at 9 brains. 10 I keep imagining Young Frankenstein, A B 11 normal. They actually started dying brains. I 12 guess that dying revealed the connection within the 13 neurons of the brain. 14 So this guy, Alexander Bain came up and said, 15 in his book on mind and body said for every act of 16 memory, every exercise of bodily aptitude, every 17 habit of recollection, train of ideas; there's a 18 specific grouping of coordination, sensation of 19 movements by virtue of specific growths in cells 20 and junctions. 21 I think he was the first one actually to start 22 putting together some sort of network diagram of 23 how we imagine the brain would work. 24 You'll see later on when we get to the kind of 25 classic neural network or algorithm, this was done 16 1 SCIENTIFIC BRIEFING 2 almost 100 years earlier. Here, you know, there is 3 this sense of grouping movement within the neuron 4 connections through there. 5 I think what happened with Bain too is he 6 started thinking about just how the human mind 7 works and how large that sort of network would have 8 to be to represent everything we're processing all 9 the time. 10 I think he found it to be overwhelming. Now, 11 we didn't just leave it up to the Brits. America 12 actually came in, weighed in -- William James, I 13 think he was fairly unaware of Bain's work. He was 14 the philosopher and brother of Henry James. 15 William James' line I know, he's at a cocktail 16 party and somebody came up to him, being a 17 philosopher said, James, is life worth living? He 18 took a drink and said, depends on the liver. 19 Anyway, I guess this is the first approach of 20 the, I call it the use or lose it theorem. When 21 the two elementary brain processes have been active 22 together, one of them on reoccurring tends to 23 propagate its excitement to the other. 24 So you actually get this notion of neuron 25 exciting another neuron. As that's done again, 17 1 SCIENTIFIC BRIEFING 2 that's some strengthening of that connection. As 3 we'll see later on, that concept is fundamental in 4 the use of a neural networks. 5 Actually, I should mention too they always say 6 artificial neural networks, you say A N N. I guess 7 because this one is the natural neural network and 8 everything else is artificial. 9 McCulloch and Pitts, trying to fast forward 10 a little bit, this one is in '43, they did some 11 sort of work between James and here. They actually 12 started building simple electric circuits that 13 would process neuron kind of junction here, and get 14 some results, actually use the logic thresholds, 15 again, kind of elementary sort of connections 16 trying to imitate the functioning of the brain. 17 Widrow and Hoff, the only reason I mention 18 this, this is 1959, and what they're looking at is 19 the binary signals in the telephone lines was quite 20 noisy. So this was actually one of the first 21 practical applications to clean up the noise in 22 telephone lines, where it actually would predict 23 the binary signal to come up. You could clean it 24 up that way. 25 Here you see their little network here. You 18 1 SCIENTIFIC BRIEFING 2 start seeing they're assigning weights. There is 3 some sort of structure here to take the error 4 in the prediction, bring it back, and strengthening 5 that connection by adjusting waves through here. 6 They also came with this delta rule, learning 7 rule. That learning rule is really how do you 8 adjust those weights and connection strengths 9 through there. 10 We'll see later on developing that sort of 11 error reduction is critical. But some rain came on 12 the parade. I think there is a bit of science 13 fiction, you know, during the history of machines 14 thinking like humans, and all this other sort of 15 stuff. 16 The early work was tallying this part of it. 17 Minsky and Papert came up with this book called 18 Ceptrons (phonetic). They really did a thorough 19 analysis of what had been done to date there. 20 Everything was pretty over-promised, that the 21 artificial neural networks could never do what 22 they had been touted to do. 23 I guess it was such a strong denunciation, in 24 a way, of where the state of art was that the A and 25 N research dwindled and went into hibernation. 19 1 SCIENTIFIC BRIEFING 2 Then, I guess, there's a U.S. Japanese 3 conference, and one part of it was cooperative 4 competitive neural network. Japan decided to come 5 out and really give it a good strong effort, to 6 really make it the fifth generation of computing. 7 I'm not a computer scientist. I can't 8 remember what the first four generations were, but 9 wires was the first and transistors, solid state 10 circuitry, I think, through there. 11 I guess the U.S. was really worried it was 12 going to get left behind. Think back to the 13 early 80s too, that's when the Japanese economy was 14 extremely strong. Weren't they buying Manhattan, 15 Pebble Beach, maybe Augusta. 16 People started getting scared, and actually 17 there was a kind of counter U.S. effort that some 18 of the materials used now were coming up. Another 19 strong thing was how do we get the learning 20 algorithm to be more robust and faster. 21 So as you train a network, you have the 22 strength and connections, how do you do that in 23 a good, robust way. 24 What happened is even though these three folks 25 are supposedly working independently, that sort of 20 1 SCIENTIFIC BRIEFING 2 thing, this guy Rumelhart gets all the credit for 3 this. It's a back propagation network. 4 You can see, this is just recent past of '86. 5 It was an error minimization approach to propagate 6 back over the network that worked fairly fast. so 7 all of a sudden you could take one of these 8 networks, train it on the million data points, 9 and it actually would work fairly quickly. 10 At the same time, '86 is when the computers 11 are starting to speed up. So all of sudden the 12 computing capacity is there to be able to handle 13 error minimization. 14 I think this is what really started opening up 15 using artificial neural networks for a variety of 16 fields. 17 Here's the classic network that you'll see all 18 the time. With these neural networks, there's 19 always the George Jetson side of this thing. It's 20 really just mapping inputs to outputs, the same way 21 a regression equation, you know, Y equal MX plus 22 B, taking X and being able to predict Y. 23 So we're still just having input to the 24 mapping to outputs. Here we have vectors coming 25 in. You can think of this just as a time series of 21 1 SCIENTIFIC BRIEFING 2 incrementing time through here. 3 This comes through in here and all gets 4 normalized in a range like zero to one, so data 5 with large ranges don't get more favoritism than 6 data with small ranges. 7 Then the hidden connections, the connections 8 to the hidden layer, and these are where the 9 weights are. So there's weights, you can think of 10 it similar to the coefficient and regression 11 weights here, but these are weights. 12 They go into the hidden layer and also to the 13 outer layer. And then the output comes out through 14 here. Between the hidden layer and the output 15 layer, there's also another transfer function, 16 usually some sort of sigmoid, you know, 17 hypertangent function through here, which is 18 non-linear. 19 This is what gives it the non-linear power. 20 It's not just with all these connections, which 21 just the connections itself, it would be fairly 22 pseudo non-linear without the sigmoid function, but 23 with the sigmoid function, it all of sudden really 24 has the non-linear power through there predicting 25 the output. 22 1 SCIENTIFIC BRIEFING 2 I'm going to try to actually bring this down 3 to real time -- oh, first neural networks learn the 4 easy stuff first. You give it set of data. You 5 can think of it almost like pattern recognition. 6 It understands that dates and makes output. 7 There's a difference also between memorization 8 and generalization. You can create these things 9 just to memorize, so given the same input, it 10 memorizes what's going on. 11 You really want it to generalize, seeing new 12 information, that it can make a generalized 13 prediction. 14 Well, I guess the military, they're intrigued 15 with the idea of using these. I guess this is in 16 the late 80s, I guess one concern they had was 17 often times you would have woods, and tanks in the 18 woods. 19 You wouldn't know if the tanks are in the 20 woods or not. Sometimes there would be woods and 21 no tanks. Maybe we can artificial neural networks 22 to envision training to be able to predict if there 23 are tanks in the woods or not. 24 So they went out and took 200 photographs, 100 25 with just woods, 100 with woods with tanks. They 23 1 SCIENTIFIC BRIEFING 2 trained the model, showed it 100 pictures, you 3 know, trained up, did well. They showed it the 4 other 100 pictures, it nailed it. 5 So everyone's excited. They say, let's test 6 it again. They went out, took another 100 7 photographs, showed the neural network. It didn't 8 get one right. 9 What happened is when they took the pictures 10 of the woods, and the pictures of the woods with 11 tanks, they took the pictures of the woods on a 12 sunny day. They took the picture of the woods with 13 tanks on a cloudy day. 14 So all the neural network did was pick up on 15 what was sunny and what was cloudy. 16 So I guess a large lesson was learned with a 17 lot of federal dollars. But neural networks are 18 pretty prevalent these days. 19 I think a lot of times -- I know in marketing 20 it's used a lot. I know my wife, one Christmas, 21 I think she probably stopped at the Starbuck's for 22 a grande or something and decided to do all her 23 Christmas shopping in an hour or something. 24 By the time she got back to the house, there 25 was already a phone call about the use of the 24 1 SCIENTIFIC BRIEFING 2 credit card was beyond normal behavior through 3 there. 4 I'm sure that's just a neural network looking 5 over the data, throwing something out where the 6 behavior pattern isn't normal. I've also heard 7 that, you know, department stores all tie together 8 so well now that Target can be running a special, 9 have a sign up somewhere, and all of a sudden 10 there's a sales drop on the special. They can get 11 a phone call from Minneapolis saying you better 12 check aisle four. See if that sign is still up. 13 The sign may have fallen down, people don't see the 14 special all of a sudden. 15 With the stock market, I mean, computer chip 16 design -- I was in a conference and actually heard 17 a guy give a talk about using artificial neural 18 networks for inventorying a Home Depot, and 19 actually imagined every shopping cart, the 20 different vectors of data points, so it's pretty 21 prevalent these days. 22 I think it's just starting to come into the 23 environmental world. What I want to do before we 24 get into the coastal situation is just talk about 25 the simpler application. This is up in Columbia, 25 1 SCIENTIFIC BRIEFING 2 backyard. South of Columbia is the Congaree Swamp 3 National Park. 4 It's one of the last bottom, you know, land of 5 hardwoods. I guess all these guys can probably 6 tell us about the glories of the place, but one 7 issue down here with the swamp and the flooding and 8 the water levels in the swamp is that you have the 9 Saluda River which is controlled by releases from 10 Lake Murray. 11 So you have a peaking station there affecting 12 the water levels down here. You have the Broad 13 River, which has kind of a low head hydro power all 14 going through down here. So -- well, what's the 15 relative contribution of these two on Congaree 16 Swamp, if you were able to control these better or 17 manage them in a more optimal way. 18 I'm not going to get into the results of that. 19 I wanted to just show kind of the power of the 20 non-linear approach. We're about to step into the 21 imagining of things and three dimensions too, so 22 warn you all on that. 23 One thing, you start thinking kind of in 24 geometry. A lot of times you look at an XY plot 25 and it can reveal a lot. What we know in these 26 1 SCIENTIFIC BRIEFING 2 natural systems, a lot of variables are interacting 3 at the same time. 4 So if you can start seeing two variables 5 interacting on a third, you can really start seeing 6 what these interrelationships are. If we can 7 imagine in the four and five dimensions, we would 8 be even that much better, but I haven't got that 9 far yet. 10 With regression, a lot of times, there's like 11 a flat plane, you've seen like a pane of glass, 12 trying to fit data. With these neural networks, 13 you actually can have curved planes, you know, 14 hyperplanes, to try to fit the data. 15 Most natural systems really have these 16 non-linear behaviors. I think in the past what 17 we've done is taken the non-linear behavior of 18 natural systems and linearized it by doing some 19 sort of log transfers, something like that. 20 Here's just the flows from the Broad River and 21 the Saluda River. This is over like two years, 22 three years. I just fit a straight line through 23 it, and you can see this fits by hand with a paint 24 brush. 25 It's not so scientific, but you can see just, 27 1 SCIENTIFIC BRIEFING 2 you know, here's the straight line fit and here's 3 where the actual behavior of the data is. 4 Here on the Saluda River, they're very 5 similar. The flows here are higher than the 6 tangent scale, but you can see the same sort of 7 behavior through there. 8 Then we kind of looked at that -- you try to 9 look at this in three dimensions. Here's the Broad 10 River and Saluda River on those water levels down 11 the swamp. Here's the two dimensional one. You 12 can kind of see that same behavior here, there and 13 this is the Saluda on the other side. This is just 14 that same image that's non-rotating. 15 You can see if you wanted to build a 16 simulation model of this, you don't want put a flat 17 pane of glass through this thing. You want to do 18 something that has kind of non-linear behavior. 19 So fitting a neural network model through 20 there, you can see this is the Saluda flows and 21 Broad River flows and the swamp gauge heights 22 through here. 23 This becomes a representation of three years 24 of data and that cloud through there. You can 25 start seeing what sort of relative contributions 28 1 SCIENTIFIC BRIEFING 2 you have, given increasing flow levels here with 3 gauge height changes in the swamp with the Saluda 4 and Broad River through there. 5 They will kind of capture that behavior of the 6 data. I took the same data and actually did just a 7 linear regression of it, and then plotted that at 8 three space here. You can see this is the plane of 9 glass doing the best it could trying to fit the 10 data, but if you are using this to try to 11 understand the system, how much you would be 12 missing through there. 13 If we move down to coastal systems, you can 14 see we all are dealing with a very important 15 coastal system, and here the interactions are just 16 tremendous. We have the water levels, tidal 17 ranges, currents, salinity, temperatures, dissolved 18 oxygen, point and non-point source systems, all of 19 these that may occur on the same time scale, may 20 not in space and time, and they all have this 21 periodic and chaotic behavior through there. 22 On top of that, they're all highly correlate, 23 and what the high correlation means is that they 24 share a lot of information content. So if you are 25 to look at water levels, especially tidal ranges, 29 1 SCIENTIFIC BRIEFING 2 we know that drives a lot of things. 3 Well, those tide ranges share a lot with that 4 water level, so you want to make sure that as 5 you're trying to understand the relative impacts of 6 these, that the information content is unique. 7 So what happens then, it becomes a variable 8 exercise in pulling out that unique information 9 through all those different time series through 10 there. 11 Just a little review on chaos theory, people 12 get it confused a lot with the childhood chaos 13 theory versus control, but it's really more than 14 that. 15 Here's a -- conductance and flows at 16 Limehouse. This -- I'm sure we all do this, have 17 our favorite plots for the year, this would be 18 mine. This shows flows and conductance system 1994 19 through 2002. So here's flows in Clyo through 20 here, here's specific conductance, think of this 21 as salt at Limehouse. 22 What we can see is kind of how those normal 23 behaviors, whatever normal is, before the El Nino 24 event in 1998. After 1998, we know what happened. 25 We had five years of drought. So you can see these 30 1 SCIENTIFIC BRIEFING 2 flow levels slowly start going down. 3 One definition of chaos is a system that is 4 highly susceptible to small changes in boundary 5 conditions. So boundary conditions are, you know, 6 occurrences are kind of happening away from the 7 system, and those things get changed just slightly, 8 and you have a large reaction or large response in 9 the system. 10 What we see here, actually, in a normal 11 period, you can see low flows going on and boom 12 they get low. All of a sudden they get a little 13 bit low. What happens to the conductance. It just 14 climbs. 15 Previous -- prior to the El Nino event, 16 conductances might get up to the 2,000 range. Drop 17 those flows just a little bit, all of a sudden 18 they're up to the 12,000 range. 19 Kind of this is the classic kind of 20 environmental chaotic signal. So the trick then is 21 how do you capture that information; in other 22 words, you look at the previous hour, you might not 23 get this. We can see through here, it's not just 24 this low flow. We had a low flow here, but it's a 25 sustained low flow, almost like the anesthesia 31 1 SCIENTIFIC BRIEFING 2 condition comes in to play a part. 3 So how do actually characterize that with the 4 data you have, to be able to get those simulations 5 to be able to match the responses we're seeing. I 6 just love to torture people with three dimensional 7 plots. 8 What we have in this direction is flows from 9 Clyo. Here is the tidal ranges down towards Ft. 10 Pulaski, and here is the conductance so this 11 becomes -- actually, this is a three, four year 12 scale. 13 That's a representation of three or four years 14 of data there. We see the same thing in the 15 previous plot. Here is the low flows here. Here 16 is the medium flows through here. As flows drop, 17 what happens -- boom. Conductivities go up the 18 wall, just shoot straight up. 19 Here you can also see, well, how does that 20 compare with the water levels. Here we have spring 21 tides and here we have neap tides. During the 22 spring tides and extreme low flows, that data with 23 conductivities will drive up there. 24 With the model what you want to do is to be 25 able to capture that -- capture that same behavior. 32 1 SCIENTIFIC BRIEFING 2 So the task then, actually, we did this, part of 3 Wiley Kitchens' work down in the swamp there, Wiley 4 is working out there and starts realizing he's 5 getting a baseline study during an extreme event. 6 What he really wanted is to be able to 7 normalize that data. He's out there and said, 8 well, what would the water levels and 9 conductivities be if it was a normal year and not 10 extreme drought. 11 That's really what we're aiming to do, so what 12 we're going to be able to do is take the flows in 13 Clyo and the water levels down at Ft. Pulaski and 14 predict salt and water levels down through here. 15 As a modeler, I would want conductivity on the 16 boundaries, but you don't always get what you want, 17 the Rolling Stones song says. We said, well, let's 18 go ahead and try with just water levels and flow 19 and see how well we can predict the conductivities 20 through here. 21 And actually, what we found out is that we 22 were able to do it well. Now again, my favorite 23 plot, but what we were able to do with just those 24 two inputs, to kind of extract as much information 25 about the low flow and sustained low flow 33 1 SCIENTIFIC BRIEFING 2 conditions. 3 We can actually take the flows up at Clyo, 4 they don't really impact down in the refuge area 5 for two days, so we can take a two day lag of those 6 flows and we can also take a 14 day running 7 average, which really brings in information of 8 well, if they were low, would it be low for 14 9 days or not, not just a spike where it just drops 10 down for a day, but really what are the conditions 11 over 14 days. 12 Then we're able to take two, 16, and 30 day 13 differences in the 14 day moving average. So you 14 can imagine like an average 14 day flow, and then 15 say, well, how was that 14 days ago, or two days 16 ago, or 16 days, 30 days. So you kind of get this 17 long-term movement of whether the whole system is 18 going down in flow conditions or up in flow 19 conditions. 20 And then down at Ft. Pulaski you get the tidal 21 signal. We have obviously the hourly water level 22 but also the 24 hour tidal ranges through there and 23 lag differences of that. 24 So then we have information of what part of 25 the tide you're in and whether you're moving into 34 1 SCIENTIFIC BRIEFING 2 and out of spring neap cycles through there. Also 3 an interesting thing with water levels, if you look 4 at the time definitive or time of day, you can see 5 the water levels moving all over the place, 6 compared with the really periodic tide range 7 through there. 8 So even though they're related, there is some 9 sort of independence there. What we're able to do 10 is take two time series, but extract a lot more 11 information out of those to build a model with. 12 Just kind of a quick summary here of a model, 13 this is up in -- this is up at 95, so the water 14 level models, we've got an R squared, we've got .95 15 which means using those inputs, we were able to 16 explain 95% of the variability of the water levels 17 up there. And with the conductivities, we're able 18 to get R squared to .85, so we were able to capture 19 variabilities going on. 20 I think this is the last 3 D plot. So one 21 thing we wanted to look at was, okay, how do these 22 conductivities respond during neap and spring 23 tides, and when do we get that real large intrusion 24 going on. 25 So the response on the left is neap tides, and 35 1 SCIENTIFIC BRIEFING 2 on B is spring tides. The normal intrusion on the 3 system happens during neap tides. You don't have 4 as much energy in the system, you kind of get the 5 intrusion going on. You can kind of see that. 6 Here is water levels going into the stream 7 here and the flows on the Savannah this way, so as 8 the flows decrease, you're going to start seeing 9 the normal intrusion that's going on. 10 Well, you can see under similar flows here, 11 that's hardly any intrusion going on, but during 12 extreme drought, you get those flows dropping just 13 a little bit, here it is climbing up the wall, as 14 we saw on the other plot. All of a sudden you're 15 getting conductivities. We've moved up to I-95, 16 but up to 400, where under kind of normal 17 conditions this might be a 140. 18 So what good is all this. I'll have Larry 19 answer that. So what we're able to do then is 20 actually build a model to predict these intrusions. 21 The question is, can we normalize conditions from 22 the drought to normal conditions. And here we can 23 see how well it's predicting intrusions going on at 24 95. 25 You can see it does a pretty good job. We 36 1 SCIENTIFIC BRIEFING 2 have kind of intrusions going on in every other 3 cycle. It's actually able to pick up the range 4 well. So knowing -- being a modeller, you've got 5 this thing a modeller has to do behind the curtain, 6 don't disturb them, those sort of things. 7 We decided to open up the dissemination of the 8 model by putting it in Excel. That way we can 9 share the pain together, you know, give it to 10 somebody else. Here we actually built a simulator, 11 with the data in Excel, where here you can -- 12 control simulation is through here, upper panel 13 control, start and stop date, time stamp, and then 14 here is our earlier version. 15 We have a newer version. You can control 16 either putting a constant flow in Clyo or some 17 percentage of historical flows. So this is the way 18 Wiley can play his games. Well, if there is 25% 19 more flow in the system, what would the 20 conductivities have done. 21 That's the simulation I did here. Here's just 22 the input data here, here's the water levels at 23 Clyo -- I mean water levels at Ft. Pulaski. Down 24 below are the flows at Clyo. The little dash line 25 is where I gave it 25% more flow. That's kind of 37 1 SCIENTIFIC BRIEFING 2 user defined. 3 And then here's the graphical display. What 4 we have is -- let me find a good one. I need to 5 edit my graph a little bit. Here's a red solid 6 line. Solid lines are the actual data. The solid 7 lines with the dot is actually the model 8 predictions. 9 You always can see how well the model did 10 compared to the actual data. So when you look 11 through here, you can see an area here that didn't 12 quite get that peak, but here it is getting it 13 pretty well. 14 Then the dash is what's the response, the 15 conductivity salinity response due to increasing 16 the flow by 25%. So with 25% more flow coming 17 down, expect that the salt would be pushed farther 18 out, and that's what we see through here. See the 19 lower line here is the response by the user-defined 20 flows, and this gray line is the difference between 21 the actual predicted conditions and the user. 22 So then you can run the long-term simulations 23 and see how do the systems behave under different 24 hydrological conditions. So again, I just played 25 that game, looking at one period through here, 2000 38 1 SCIENTIFIC BRIEFING 2 intrusion, and here again is the 25% increase in 3 flows over that period. 4 And then the actual data is in red, so you can 5 see this is the intrusion going on at 95, and then 6 with 25% more flow is blue. So you can see how 7 much is knocked down in peak conductivities of 8 2,500. All of sudden it gets knocked down below 9 500. 10 So that -- that work was done using the USGS 11 river network and went through 2000. We used the 12 data between 1994 and 2000. Subsequent to that, 13 the kind Georgia Ports were looking at actually 14 tying the same kinds of network technology from the 15 model, 3-D model that's being used, to the marsh 16 response and impacts. 17 So here GPA and USGS got a lot of time series 18 data in the marshes themselves. It's taking the 19 same kind of response, but not in the river, but 20 translating that up to the marsh. It's using the 21 same data mining neural network applications to do 22 that. So again, it would be building a simulator 23 similar to that. 24 Excel 1, though, I can actually take the 3-D 25 hydro out and put -- and be able to predict 39 1 SCIENTIFIC BRIEFING 2 response in the marshes themselves. Here's the GPA 3 network through there that ATM did sampling, two 4 years of data on there. Here's the USGS set that 5 Wiley Kitchens was sampling. 6 We have two years of high frequency time 7 series through there. We'll be using that data in 8 the same way to be able to determine the responses 9 through there. 10 This is the approach, the first thing we're 11 going to do is the river network itself, including 12 the 2001 and 2002 data through there. We've done 13 that and I think it's really improved the models, 14 because we were that extreme where we're climbing 15 that wall through there, so we're getting a whole 16 range of behaviors through there. 17 So we really have been able to predict the 18 tidal intrusions as well. Then we're going to 19 integrate the GPA stations from 1997 and 1999 into 20 that network to kind of get a better definition 21 throughout, you know, the wildlife area there of 22 responses. And finally, integrate that to the USGS 23 GPA marsh stations. 24 So again, we have a whole network where we can 25 look at flows from Clyo, the water level at Ft. 40 1 BRIEFING QUESTIONS 2 Pulaski, and get the marsh response. Once that's 3 done -- we also do some visualization, 2-D 4 visualization planned view of seeing how the 5 results move around. 6 Then we can take the model predictions from 7 the 3-D model and put that in there and translate 8 those out to the marsh. We'll be able to get a 9 pretty good, high fidelity prediction of what the 10 response in the marsh is using the predictions from 11 the 3-D model. 12 So we'll synthesize all the time series data 13 collection that's going on with the 3-D modelling 14 effort, using the same data mining techniques used 15 in the earlier studies, combined with 2-D 16 visualization, other response through there, and 17 the output would be imported into the marsh 18 secession model and the GIS application in the 19 model, putting out X Y Z data. And with that, I'm 20 done. I was very appreciative there was no time 21 frame on the agenda there. 22 MR. DYSART: Questions? 23 MR. WRIGHT: When and how do you develop the 24 criteria for testing the performance of the model? 25 MR. CONRADS: That's a good question because 41 1 BRIEFING QUESTIONS 2 one thing with a neural networks is what you are 3 doing is you're fitting surfaces in space, and the 4 same power you get that is flexible also means that 5 you can over-train, over-predict, the same way like 6 with if you're trying to do join some lines -- if 7 you're fitting points in a line with polynomials, 8 some sort of curved line, and you want to hit every 9 one of those things with a higher order one, all of 10 a sudden, that thing can start looking like rope 11 all over the place. 12 You might be hitting those points, but in 13 between it's going crazy. The same thing can 14 happen with a neural network model. I got some 15 great examples where you have notions of physics 16 that probably shouldn't be violated, like a water 17 quality model that you increase the BOD loading, 18 you shouldn't get an increase in the water quality 19 DO level. We have one model, whew, all they needed 20 to do was put more BOD in that thing. 21 So you really have to be careful of 22 evaluating them. We have the visualization 23 routines that look at the surfaces, all the 24 different combination of variables. 25 Also one of my favorite word, bifurcating the 42 1 BRIEFING QUESTIONS 2 data, which is just splitting your data sets. You 3 split your data sets into the training and testing 4 data set. 5 If you are real lucky and have a lot of data 6 you can actually have another kind of testing set 7 to make sure it's fitting everything correctly. 8 Keith. 9 MR. PARSONS: Keith Parsons, Georgia DNR. All 10 this data that you have compiled and these models 11 you've run have been based on a known channel 12 depth. 13 MR. CONRADS: Right. 14 MR. PARSONS: Is there anything in the model 15 that's going to be able to predict how the model is 16 going to function at different depths now? 17 MR. CONRADS: What we'll do there later, 18 that's why we're not doing the deepening, we'll be 19 taking the output from H and S model and using that 20 to make the predictions with the changes in the 21 geometry and the neural network data mining model 22 will then translate that into the model 23 MR. PARSONS: So you can get a model based on 24 deeper channel depths? 25 MR. CONRADS: Yeah. That's how the whole 43 1 BRIEFING QUESTIONS 2 thing was designed is to be able to take that 3 output and then translate that into the marsh. 4 In other words, that connection between the 5 river and the marsh, there's really -- the physics 6 aren't there to explain how that happens. Here 7 we're lucky we have a wealth of data there. We can 8 say okay, let's let the data show what correlation 9 is. We can let the 3-D model make the predictions 10 of the change of water level and salinity through 11 the deepening, and use that output as input into 12 this model. 13 MR. PARSONS: Is there any data available, 14 prior to the previous channel deepening? I believe 15 it went from 38 to 42 feet in early 90s. 16 MR. KEEGAN: There's flow and water level 17 data. I think that's, in large part, what you 18 guys -- 19 MR. CONRADS: Yes. The GS network was in 20 different stages all through these periods there. 21 I mean, there is data of previous channel changes, 22 so you could go back to that and say okay, at this 23 channel depth, parameterize that depth with 24 everything else, but that seemed fairly ambitious. 25 There are folks who probably say you can 44 1 BRIEFING QUESTIONS 2 parameterize all the channel changes too and try to 3 predict with data mining, but I think it's 4 something that with this study, you need to use 5 that 3-D model to do those changes. 6 MR. PARSONS: Okay. Thank you. 7 MR. DYSART: David. 8 MR. KYLER: As powerful and as interesting as 9 this may be for hydrologic or some water chemistry 10 modelling predictions, it seems to me what's of 11 concern to many of us around the table here has to 12 do with biological functions that correlate with, 13 or are caused by, or have some unknown thresholds 14 of effect, which there's no precedent for, at least 15 there's no data for serving as a predicting basis 16 for anticipating the effects of the hydrological 17 chemical changes. 18 Likewise, there's probably some things in the 19 biological world that shift from being a 20 correlation with physical and chemical changes to 21 actually becoming causalities. I was thinking of 22 wisteria, which remains relatively benign and a 23 passive thing until a certain condition is reached, 24 and it becomes an aggressive, opportunistic, 25 destructive organism. Any comments on that kind of 45 1 BRIEFING QUESTIONS 2 stuff? 3 MR. CONRADS: Well, yeah. You know, I think 4 one thing with models, you always get concerned 5 about interpolation and extrapolation. So I think 6 what your question is getting at, are you starting 7 to extrapolate beyond conditions never seen before. 8 To me, that's a concern with whatever approach 9 you take. I think a lot of times we push models 10 way beyond the range of conditions they have been 11 calibrated under. 12 The power here is in having the data, and the 13 fuller the range of data you have to show the 14 behavior of a system, that's where the strength is. 15 If you only have three or four data points, I 16 don't know how much more this method can reveal 17 from those data points there. You know, if it's a 18 concern with wisteria kind of lurking in the 19 background, you have to have some major, you know, 20 response going on to get it to flourish that you 21 haven't seen in the system before. I mean, that's 22 anybody's game to try to predict that. 23 MR. KYLER: Let me see if I'm drawing the 24 right conclusion. If so, I want to emphasize that. 25 I don't think that was made clearer in what you 46 1 BRIEFING QUESTIONS 2 said. 3 The nature of these models is physical and 4 chemical, but not biological, mostly because we 5 don't have much historic data or understanding of 6 these relationships in a biological system. 7 MR. CONRADS: Right. That's not to say you 8 didn't have good biological data. That's something 9 that could be done. I don't know what sort of 10 biological data is out there. 11 MR. KYLER: Sure. 12 MR. EUDALY: If I could add a little bit, and 13 I don't know how familiar you are with the other 14 studies being done, we've talked about some, but 15 the marsh secession model, which this information 16 will support is the get a biological prediction. 17 We've got the data from the marsh, a couple years 18 of data related to water chemistry, salinity, soil 19 conditions, various factors, and a model has been 20 created which I'm sure will have a presentation 21 probably fairly quickly in this group, which 22 relates the marsh composition, species composition 23 and community, to those physical and chemical 24 factors. 25 This effort is to, first of all, we have the 47 1 BRIEFING QUESTIONS 2 prediction from the physics-based salinity model. 3 Then this would link that prediction to what 4 happens to the physical chemical factors in the 5 marsh. Then the structural equation model or the 6 marsh secession model will take that information 7 and predict what will change, over the surface of 8 the marsh, in terms of the vegetation community. 9 So it's the whole group of studies will be 10 integrated and try to predict -- I think what you 11 are getting at -- the effect on the marsh itself. 12 MR. KYLER: Well, that's not the whole 13 biological world -- 14 MR. EUDALY: No. 15 MR. KYLER: -- as potentially affected by the 16 project. It's a component. The other thing is if 17 you are just talking about the ecosystem of the 18 marsh is characterized by certain plant species 19 that may or may not say anything about effect on 20 innumerable animal species at various stages of 21 their lives, habitat, spawning, feeding, all that 22 kind of stuff. 23 So I'm not sure how much can be inferred from 24 a broader array of biological -- potential 25 biological causalities, even if you have a fairly 48 1 BRIEFING QUESTIONS 2 accurate way of monitoring the marsh. Also, it 3 doesn't speak to unprecedented threshold effects, 4 which there is no historical example of observing 5 and therefore predicting. 6 MR. EUDALY: Well, we tried to -- going into 7 this, we tried to pick out the sort of key 8 resources in the marsh, you know, low salinity 9 marsh is a key resource, especially on the Savannah 10 National Wildlife Refuge. 11 We do have some data of the marsh composition 12 to animal use, so forth. So it was an effort to 13 try and pick out a key resource to do evaluation on 14 that. You obviously can't evaluate the impact on 15 everything, so you have to make some decisions on 16 priorities. 17 MR. KYLER: I don't mean to be criticizing. 18 I'm just saying, I think we need to be clear about 19 the limits of that model even with predicting the 20 marsh impacts. 21 MR. EUDALY: Oh, I agree, and you need to 22 understand the assumptions that go into it. When 23 you start linking a lot of models, there is some 24 room for error there. It's just a matter of 25 trying to do the best we can in predicting the 49 1 BRIEFING QUESTIONS 2 impacts. 3 MR. DYSART: Teri. 4 MS. LEFFEK: I noticed on one of the slides 5 you had R squared .95 and the other was .85. 6 What's an acceptable R squared for a model, or is 7 that dependent on each different model. 8 MR. CONRADS: I think it depends on what you 9 are looking for because there are times where what 10 we do is break down signals. One good example I 11 have, let's say it was a DO model doing DO 12 modelling using the same technique. 13 If you predict DO, which is the temperature, 14 you can get an R squared of .85. Well, anybody 15 with a deterministic model, they got an R squared 16 of .85, and it keeps jumping up and down saying I'm 17 down, you know, that's it. But what happens is 18 that being able to predict it with temperature 19 probably isn't what you are interested in. 20 You want to see what the impacts of point the 21 sources or non-point sources are. Then you're 22 looking at that last 15% of variability, so in a 23 case like that, we turn around and actually use 24 percent saturation, and then we can look at BOD 25 loading. You might get an R squared of .3. 50 1 BRIEFING QUESTIONS 2 Normally, you go oh my god. I'm going to 3 be fired, but then you say well, you know, of that 4 remaining variability 30% of it can be attributable 5 to this one variable. 6 Then ar R square of .3, you're excited about, 7 so it's hard to put a criteria on that, you know, 8 but .85 is doing pretty well. 9 MR. DYSART: Would you clarify, who are you 10 working with on this, who are you interfacing with? 11 MR. CONRADS: Well, okay. Actually, I should 12 mention my co-conspirator on this Ed Roehl with 13 Advanced Data Mining up in Greenville, South 14 Carolina. 15 He and I have been collaborating for about 16 five or six years on this technique. Actually, we 17 have a cooperative research and development 18 agreement with USGS and his firm. It's part of, I 19 guess, the federal government's technical outreach 20 of trying to merge private sector technology and 21 public sector need. 22 That's how we've teamed up. We're doing this 23 work and doing it in support of the marsh studies 24 that are going on of ATM, John Bossart ,and Wiley 25 Kitchens as a coop unit down in Gainesville. 51 1 BRIEFING QUESTIONS 2 MR. DYSART: More questions. Judy. 3 MS. JENNINGS: Paul, clearly I'm interested in 4 how it relates to this project, but I'm still back 5 on A and N. That's a fascinating concept. Once 6 you understand a sigmoid function, even though it's 7 non-linear, once you understand it, it can still be 8 predicted? 9 MR. CONRADS: Right, right. It has that 10 non-linear form in there, so it can translate that 11 into non-linear behavior of the data. 12 MS. JENNINGS: I guess the same question for a 13 chaotic pattern, as long as you understand the 14 chaotic nature of it, it's also predictive? 15 MR. CONRADS: Right. And there the 16 predictions are on a shorter horizon, but by being 17 able to pick out the chaotic, what's left is 18 signal, hopefully you get the noise out, then you 19 have a really periodic signal that you can 20 predict well. 21 Then hit it with the chaotic, then you've 22 built the signal up. That's where a lot of the 23 power is in the data mining, because it's breaking 24 the signals apart. Then the neural network is 25 being able to predict signals building back up 52 1 BRIEFING QUESTIONS 2 trying to reconstruct the way we wanted the data. 3 MS. JENNINGS: I guess I was thinking chaos as 4 in a system being highly susceptible to changes. 5 MR. CONRADS: Right. 6 MS. JENNINGS: Do you really want to take the 7 chaos out? 8 MR. CONRADS: Well, to understand it you have 9 to take the chaos out. By removing it, you can 10 really see it, and say okay, what do I need to be 11 able to predict this. 12 The process is like breaking something apart, 13 like taking a washing machine apart to understand 14 each piece and then assembling it back together to 15 get as good a prediction as you can. 16 MR. DYSART: Chris. 17 MR. SCHUBERTH: As good a prediction as you 18 can, that's -- that's what troubles me a little 19 bit. When all of the calculations and everything 20 you've described here is done, and you throw the 21 switch, I just think of the meander loops in the 22 North Platte River, always marvel at the fact the 23 that the meanders loops don't seem to follow any 24 particular form. 25 When the switch is thrown, everything that is 53 1 BRIEFING QUESTIONS 2 the modelled to predict will come to pass in the 3 natural system, when it operates. 4 MR. CONRADS: Oh, I'd love to be able to say 5 that. 6 MR. SCHUBERTH: Okay. 7 MR. CONRADS: But you know, I mean, it depends 8 on the behavior of the system. We use models in a 9 lot of different ways. Just a simple rating curve 10 relating stage and discharge in a river is a model. 11 So we have rating curves on rock formations 12 where we probably developed that curve 50 years ago 13 and it's probably holding up quite well. You go to 14 a sandy channel that's meandering, you have to 15 develop that thing every six months. 16 It really depends on the system. When I first 17 got into this, gosh, this is great, if you could 18 find some sort of generic approach to take, but 19 these natural systems all have their kind of unique 20 behaviors. You have to capture it. If you don't 21 have it in the data, then it's very difficult to 22 project out there. 23 MR. SCHUBERTH: Thanks. 24 MR. CONRADS: That's one way we're real 25 fortunate here. We have the El Nino event to 54 1 BRIEFING QUESTIONS 2 extreme drought, as far as trying to understand the 3 behavior with water levels and flows on the water 4 levels and conductivities in the wildlife area. 5 So we have an incredible broad range of 6 behaviors right there. So, within those confines, 7 we probably can make fairly good predictions. We 8 get the mother of all El Nino events, then you're 9 extrapolating, and hopefully it will still be able 10 to pick that up. 11 MR. DYSART: Okay. Further questions. Larry, 12 do you have any comment about this, how this 13 interfaces with modelling. 14 MR. KEEGAN: No. I'd be afraid to even try 15 and comment on what Paul does. 16 MR. CONRADS: Larry has run the simulator. 17 MR. KEEGAN: That's true. I can attest that 18 even someone with my understanding can run the 19 product he produces. I'm not sure I know what to 20 do with it once it runs, but I can do it. 21 MR. CONRADS: We'll get to that in the next 22 stage. 23 MR. DYSART: Seeing no further questions, 24 Paul, we appreciate your briefing. It was very 25 helpful. I'm sure it answered all of Will's 55 1 GPA UPDATE 2 questions. Thank you very much. Next on the 3 schedule old business, first item is a GPA update. 4 Morgan. 5 MR. REES: Larry. 6 MR. DYSART: Larry. 7 MR. KEEGAN: Go over the status report 8 briefly, if anybody has any questions, please 9 interrupt and we'll deal with them. 10 Interagency coordination, coming up hopefully 11 is the wetland coordination effort that Bill is 12 trying to put together. Still tentative about the 13 1st of July. We're not sure of that yet. Number 14 of agencies looking to attend, and that will start 15 to -- what's the right word -- on the agenda is to 16 get all of those agencies to look at the efforts 17 that Fish and Wildlife and GPA and USGS have been 18 putting together for marsh secession modelling and 19 how it will be used, how it will work. 20 Fisheries continue -- looks like we're pretty 21 close to having the evaluation procedures 22 finalized. We're waiting to hear back from various 23 agencies. 24 Ground water I'm going to defer to Doug to 25 talk about that. That's on the agenda. Water 56 1 GPA UPDATE 2 quality, moving ahead, we had a meeting in Atlanta 3 with the water quality group, went over the data 4 characterization that's being used for the 5 dissolved oxygen model development, and a pretty 6 good discussion I thought. 7 There's a question on the table from that 8 group about the K rates, kinetic rate parameters to 9 be used, that's got to be resolved. Sediment 10 placement, no big movement on that, we're still 11 looking to figure out what all the requirements are 12 to go ahead and model before we actually start 13 modelling. 14 Plan formulation, moving slower than we'd 15 like, but we don't have the information to make it 16 move a whole lot faster, so that's okay. 17 No big change in the hydrodynamic and salinity 18 model calibration. We're awaiting acceptance of 19 resolution from ERDC and Georgia DNR. Marsh 20 secession model, you heard a lot about today. 21 There's work going on to refine some of the 22 components of that, one of which Ed Eudaly talked 23 about briefly which was a structured element 24 modelling, that actually is doing the biological 25 modelling. 57 1 GPA UPDATE 2 That's a big part of the refinement, and we're 3 looking at when that can be done. I was talking 4 with Ed about it a little bit earlier. You can see 5 the other things that will happen to refine that. 6 Ship simulation, the draft report is in, being 7 reviewed now. I know the district's going to work 8 on getting that out to the pilots' association, the 9 Coast Guard, Georgia DOT for their look. 10 Hopefully, that will be finalized pretty soon. 11 Sediment quality moving, it looks like we're 12 getting closer to an agreed upon methodology. 13 We're still looking for feedback from the agencies 14 involved, maybe by mid June. I don't know how 15 optimistic that is, but we'll see. 16 CSS Georgia, the contract has just been 17 awarded to actually do on site investigation now at 18 the Georgia wreck site. That just happened a few 19 days ago, and that's kind of exciting. That's 20 going to be, if I'm not mistaken, the first time 21 that sort of onsite investigation of that site has 22 occurred. 23 MR. PARKER: That's incorrect. Back in the 24 late 70s earlier 80s, we did a site investigation 25 Texas A & M's marine archeology department did some 58 1 STATUS COMMENTARY 2 diving on the wreck. 3 MR. KEEGAN: Okay. My mistake. I don't have 4 the details over the server, but I know this is a 5 pretty thorough effort to try and figure out where 6 to go next. 7 Any questions? This has been posted for a 8 while. Anybody have anything they want to ask 9 about? 10 MR. DYSART: Seeing no questions, we'll move 11 on. Doug, how about the status of the Aquifer 12 Committee recommendations. 13 MR. PLACHY: I'd like to have Chris in the 14 room when we do that. 15 MS. MOORER: Can I request a break? 16 MR. DYSART: How about a break -- 10 minutes. 17 (Short Break) 18 MR. DYSART: Okay. We'll reconvene the 19 meeting and the next item on the agenda was Doug 20 Plachy commenting on the Aquifer Committee 21 recommendations. Doug. 22 MR. PLACHY: Unlike probably the prior SEG, 23 this SEG I have quite a bit to go over in the 24 report which is good. 25 I'm going kind of go down through -- there's 59 1 STATUS COMMENTARY 2 so much information, if I just tried to talk off 3 the top of my head, I'd probably forget something 4 or miss something. So I'm going to go down and 5 read through this. Unfortunately, the gentleman 6 who was going to bring copies for distribution 7 still hasn't shown up this morning, so this will 8 get posted to the website. 9 I'll ask Larry to go ahead and post this. I 10 apologize, because my intention was to pass out a 11 copy of everything I'm going to try to cover this 12 morning -- part of the chaos theory, I guess. 13 The Savannah District has been studying the 14 potential impacts of the navigation channel on 15 the Floridan Aquifer for well over 20 years and 16 spearheaded the 97/98 study due to the necessity to 17 protect this natural resource. 18 The growing concern by state and federal 19 agencies regarding the vulnerability of ground- 20 water resources in the coastal area, due to overuse 21 and seawater encroachment is well known. 22 The miocene unit, a confining unit consisting 23 of low-permeability clays, clayey silts, and clayey 24 sands, overlies the Upper Floridan Aquifer. The 25 miocene unit consists of two lithologic units 60 1 STATUS COMMENTARY 2 referred to as unit A and unit B. 3 The Upper Floridan Aquifer, the most important 4 source of groundwater in the general reevaluation 5 study area, has been extensively developed in the 6 Savannah area where some 82 million gallons per day 7 of water, groundwater, are presently withdrawn from 8 this aquifer system. 9 This pumping has created a cone of depression 10 centered in Savannah, causing a downward hydraulic 11 gradient which increases the potential for vertical 12 leakage of seawater through the miocene confining 13 unit into the Floridan. 14 The U.S. Army Corps of Engineers, Savannah 15 District developed and performed a study focusing 16 on the area along the present navigation channel 17 between approximate stations 20 plus 00 and minus 18 23 plus 00, a portion of which a local structural 19 feature, the Tybee high, causes strata to be 20 elevated and thinned. 21 The principle objective of the investigation 22 was to determine if additional deepening of the 23 Savannah Harbor Channel has the potential to impact 24 aquifers, in particular the Upper Floridan Aquifer, 25 under the navigation channel. 61 1 STATUS COMMENTARY 2 Although challenged by some, the findings of 3 the 1997/1998 study are technically valid and 4 scientifically based. The hydrogeology of 5 sediments beneath the project area was evaluated 6 by a combination of extensive review of previous 7 studies, review of state and federal USGS and Corps 8 well records, eight new core borings, a sub-bottom 9 geophysical survey, analysis of groundwater from 10 the new and existing test wells, and a laboratory 11 analysis of physical properties of the miocene 12 confining unit. 13 The 1998 USACE study concluded that the 14 proposed dredging will have no noticeable effect on 15 the quality and quantity of groundwater with the 16 Upper Floridan Aquifer. 17 Although the results and conclusions from the 18 study are reasonable, additional data would better 19 support and confirm the conclusions. Better 20 mapping of the confining unit thickness, and a more 21 detailed discussion of vertical permeability 22 results in hydraulic gradients will be beneficial. 23 In addition, studies will focus on impacts to 24 chloride levels in the Upper Floridan, not on 25 whether saltwater is moving downward from the 62 1 STATUS COMMENTARY 2 surface along the river channel. 3 The key question that needs to answered in the 4 Tier II EIS is what change would be expected to 5 occur in chloride concentrations in the Upper 6 Floridan Aquifer due to project dredging. We, 7 therefore, want a well-considered evaluation of the 8 risks involved in the project. 9 The Savannah District determined that 10 supplemental data will be beneficial to augment and 11 build on the previous Savannah Harbor Expansion 12 Project study, the one conducted in 1997/1998. To 13 this end, supplemental studies including additional 14 seismic surveying, additional land and marine 15 drilling incorporating sediment pore-water analysis 16 and multi-level well technology and trial aquitard 17 testing will be performed. 18 The purpose of plan is to address the 19 following, determine what, if any, impact removal 20 of additional miocene sediments, compromising the 21 upper confining unit of the Upper Floridan Aquifer, 22 within the dredging prism will have upon the water 23 quality of the Upper Floridan Aquifer in the 24 Savannah area. 25 Determine the change in rate and quantity of 63 1 STATUS COMMENTARY 2 saltwater leakage, through the Upper Floridan 3 confining unit, that may result from harbor 4 deepening. 5 Determine if there are changes in chloride 6 concentrations, salinity, with time in the Upper 7 Floridan Aquifer that may be caused by harbor 8 deepening alternatives. 9 Determine the hydraulic properties, salinity, 10 and hydraulic head, and the spatial variability of 11 these parameters of the Upper Floridan confining 12 unit in the project area. 13 Determine the hydraulic properties and 14 geometry of various paleochannels in sediments 15 below the river channel. Better define the 16 geologic framework in the channel area. 17 The following recommended plan of study is a 18 synthesis of ideas, concepts, and viewpoints, and 19 perspectives, and positions, resulting from over a 20 three year period from 2000 through 2003. 21 There are substantial number of reviews and 22 discussions by the Savannah District 23 hydrogeologists about the conclusions of the 1998 24 report and what, if anything, could or should be 25 done to increase the scientific confidence level of 64 1 STATUS COMMENTARY 2 the conclusions documented in the report. 3 We, therefore, have the following plan that we 4 have recommended, and I'm going to go through those 5 now. One is to develop a 3-D numerical model of 6 the hydrologic system, focused on the navigation 7 channel to compare the projected various 8 combinations of hydraulic properties, and confining 9 unit thickness on chloride content in the Upper 10 Floridan Aquifer. 11 This -- and then supplemental data collection, 12 conduct additional sub-bottom seismic surveying in 13 the area of greatest concern, which approximate 14 river stations 20 plus 00 to minus 25 plus 00, with 15 particular interest in the area of paleochannels. 16 And just for a framework in your mind, that's 17 approximately where Fields Cut is between area 14 B 18 and Jones Oyster Bed Island, and then offshore 19 about three miles out of Tybee. 20 Conduct additional land and marine continuous 21 core borings, incorporating new methodologies in 22 sediment pore-water analysis and multi-level well 23 technology to further characterize the miocene 24 confining unit and in-filled sediments of 25 paleochannels. 65 1 STATUS COMMENTARY 2 Conduct trial pumping tests on two Upper 3 Floridan wells located adjacent to the Savannah 4 River channel, to determine if hydraulic testing of 5 the miocene confining unit is feasible. 6 If results of trial pumping tests indicate 7 hydraulic testing is feasible, and if such data is 8 determined to be critical, designing and conducting 9 full aquitard tests at up to five sites may be 10 necessary. 11 The modelling along with pore water and other 12 data may help determine the need for the additional 13 testing; therefore, a decision aquitard testing 14 will be delayed until at least the preliminary 15 modelling is completed. 16 Combine existing geologic, hydrogeologic, and 17 engineering data from previous studies, projects in 18 a comprehensive harbor-wide geographic information 19 system. Include -- included in the GIS will be 20 available historic data from earlier dredging 21 projects, total Upper Floridan withdrawals, and the 22 USGS, Georgia EPD, and South Carolina DHEC 23 water-level data, along with any new data, 24 including any applicable data from ongoing State of 25 Georgia Sound Science Initiative. 66 1 STATUS COMMENTARY 2 Future data will then be added to the GIS when 3 available to facilitate enhanced analysis and 4 visualization of potential aquifer impacts. 5 The 1997/98 report will be updated via a new 6 report to include the information, data, and 7 findings and/or conclusions of these supplemental 8 studies. 9 The grounds for conducting the additional 10 studies, although the results and conclusions from 11 the 1997/98 study are reasonable, additional data 12 would better support, confirm the conclusions. 13 Better mapping of the confining unit thickness 14 and a more detailed discussion of vertical 15 permeability results and hydraulic gradients would 16 be beneficial. 17 In addition, studies will focus on impacts to 18 chloride levels in the Upper Floridan, not on 19 whether saltwater is moving downward from the 20 surface along the river channel. 21 Although there are available regional data 22 pertaining to the hydraulic characteristics of the 23 miocene age upper confining unit, which is referred 24 to as the Hawthorn Group, there are relatively site 25 specific data available at this time. 67 1 STATUS COMMENTARY 2 The 97/98 study collected core samples from 3 six locations within the immediate study area and 4 submitted 22 samples from those cores for 5 laboratory analysis to determine vertical hydraulic 6 conductivity and to determine grain size fraction 7 and their statistical distribution. 8 Sub-bottom geophysical survey data is key to 9 the work necessary for a complete and comprehensive 10 EIS. The survey technique of the '98 study was 11 clearly effective at delineating the relevant 12 strata and paleochannels; however, an extensive 13 follow-up survey is warranted to better define the 14 thickness and elevation of extra and to better 15 understand the geometry and attitude of the 16 paleochannels. 17 Core borings were sufficient to obtain a range 18 of vertical permeability, however, it is 19 recommended that more borings be taken and that 20 they focus on improving the understanding of the 21 in-filled material in the relic channels. 22 The number of additional test wells will be 23 based on the need for assessing the gradient 24 between the surficial aquifer and the Upper 25 Floridan Aquifer, within the vicinity of the 68 1 2 Savannah River, and specifically the navigation 3 channel. 4 A reasonable distribution of head and chloride 5 data is needed to calibrate a 3-D model for the 6 Tier II EIS. A 3-D hydrogeologic numerical model 7 would be useful in better estimating the saltwater 8 leakage in the channel in the without project 9 condition. 10 The estimated leakage will be broken down by 11 segments of navigation channel and by paleochannel 12 and then focused on the predicted change in amount 13 due to proposed dredging. This amount will then be 14 translated in predicted changes in chloride in the 15 Upper Floridan Aquifer, which relates to the real 16 issue, water quality in the Upper Floridan Aquifer. 17 And I believe -- yeah, and that's essentially 18 what I wanted to cover this morning, which is 19 essentially specifically what the Corps has 20 determined needs to be done, and some of the 21 information behind which those conclusions were 22 made. 23 MR. DYSART: Questions, comments? 24 MR. SCHUBERTH: Doug, not reading makes it 25 difficult to integrate, but from what you read, 69 1 2 basically incorporated most of the recommendations 3 that came out of the working group, in your 4 document there, in terms of following through to 5 determine what needs to be done further? 6 MR. PLACHY: Yes. All of the concepts in 7 those recommendations have been embodied. 8 MR. SCHUBERTH: It sounded like they were. I 9 could hear the familiar words. So at this 10 juncture, this is a recommendation, so what's going 11 to happen next? 12 MR. PLACHY: These recommendations, I mean, 13 this decision and listing of work, I've given a -- 14 I forwarded basically a read-ahead copy to GPA, and 15 there's a letter being staffed at the district now 16 that will be signed out going to Mr. Schaller that 17 lists what the district has determined, and where 18 we are in regards to coordinating that decision. 19 In other words, the decision has finally 20 decided of specifically what work needs to be done, 21 but the next step, of course, is take that and vet 22 it back to South Carolina DHEC, Georgia EPD, and 23 USGS, to make sure. 24 Because what we want back from them, 25 essentially, is something in writing, they agree to 70 1 2 what we've come up with. Even though they have 3 been part of how this came about, we want to make 4 sure there's no loose ends or anything. 5 MR. SCHUBERTH: Right. 6 MR. PLACHY: So that's where we are in regards 7 to that. Then the next step would be a decision on 8 how to go about implementing and putting together 9 the specific scopes of work on how to go about 10 doing each one of those items. 11 MR. SCHUBERTH: What you just read, what do 12 you call that? You said you retracted 13 recommendation -- decision? 14 MR. PLACHY: Yeah, it's a decision. I don't 15 know what the word is -- 16 MR. SCHUBERTH: It's got a designation of 17 decision? 18 MR. PLACHY: I can't remember the right word 19 off the top of my head. It's not a recommendation. 20 It's a decision by the district engineer on what is 21 required to meet the needs of the EIS. 22 MR. SCHUBERTH: And you're going to post that 23 on the web page. 24 MR. PLACHY: I will give Larry a copy of what 25 I went over this morning electronically and he can 71 1 2 post that on the harbor site. And I believe, and 3 correct me if I'm wrong, Larry, the advanced letter 4 that went from myself to Larry that kind of 5 outlined -- that specifically outlined what needs 6 to be done with a little bit of background 7 information, I believe Larry you had posted that to 8 the harbor deepening site as part of the monthly 9 update that he went over earlier. 10 MR. KEEGAN: That's correct. 11 MR. PLACHY: Those two pieces will be out -- 12 one's been out there. 13 MR. SCHUBERTH: Where are we going to find 14 this -- on your page -- Aquifer Committee -- 15 MR. KEEGAN: The letter Doug sent me is 16 staffed -- the May status report, it is up under 17 new items posted right now. 18 MR. SCHUBERTH: Under new items. 19 MR. DYSART: Bob, did you have a comment? 20 MR. SCANLON: He answered it. 21 MR. DYSART: Any further comments -- David. 22 MR. SCHALLER: I think from our perspective, 23 the time and effort devoted to this subject is 24 indicative of the complexity of the entire subject. 25 I think it goes to the question of thoughtfulness 72 1 2 and care also devoted to this subject. 3 We did get a letter from Doug. We're going to 4 ask a few questions about that so that we can 5 better understand the determination that has been 6 made. 7 We've only had that letter for a couple of 8 days. All of us in this room, of course, many not 9 in this room, continue to be committed to the 10 premise, the fundamental rule that no harm will 11 come to the aquifer through the proposed deepening 12 project. 13 We appreciate the efforts of the working 14 group, the Aquifer Working Group, the 15 recommendations that they made, of course, I think 16 as Doug has acknowledged, were delivered to the 17 district and no doubt found to be of significant 18 value to them, as they went through the 19 deliberative process. 20 MR. DYSART: Other comments? Okay. Before we 21 go to the next item, anybody who came in and has 22 not introduced themselves, would you please do so 23 now? 24 MR. FARMER: Bill Farmer, citizen. 25 MS. COLLINS-RAHN: Lucille Collins-Rahn. 73 1 MITIGATION 2 MS. JENNINGS: Judy Jennings, Georgia Sierra. 3 MR. MOORER: I don't think I introduced 4 myself, Hope Moorer, Georgia Ports Authority. 5 MR. SCHUBERTH: Chris Schuberth, Chatham 6 Environmental Forum. 7 MR. DYSART: The next item is concerning 8 mitigation. That has been requested to be rolled 9 forward, and so there will be some continuing 10 discussion on that at each meeting. Anything 11 anybody wants to say on that? Teri. 12 MS. LEFFEK: I'd like to ask that be deferred 13 to the Operating Guidelines report. 14 MR. DYSART: Without objection, we will. Bob 15 and Chris asked that we have an item about the 16 future of the Aquifer Committee and its role in 17 conducting future commentaries. 18 MR. SCHUBERTH: I guess something has to be 19 said, but I'm not too sure what. Clearly, Doug has 20 just outlined an ambitious effort, that is an 21 outcome of the meetings that have been held by the 22 Aquifer Committee, and more importantly than those 23 meetings, the meetings of the working group of the 24 Aquifer Committee over the past -- the past two, 25 three years. 74 1 AQUIFER COMMITTEE FUTURE 2 This is where I think the working group had 3 hoped to be, at some point which is now, you know, 4 some further investigation would go forward, and 5 that's happening. 6 So the Aquifer Committee, basically, takes a 7 backseat, as far as I'm the chair and personally 8 concerned, there is no reason to meet, unless if 9 it's thought I should e-mail people, who have been 10 participating in this committee, where we are today 11 regarding this issue. 12 MR. DYSART: Bob. 13 MR. SCANLON: I see a rather simplistic 14 approach being the Aquifer Committee was disbanded, 15 the recommendation to this group, a few meetings 16 back. I just think the Aquifer Committee should be 17 reestablished, it can be just as it is still a 18 standing committee is all I would like to see, so 19 we can continue to review this work. 20 I don't see any need for meetings at this 21 time. I don't want to see the Aquifer Committee 22 disbanded. That's my only point, Chris. 23 MR. SCHUBERTH: I agree with that. I had 24 assumed, probably incorrectly, that we were still 25 viable, but I think the decision at the early SEG 75 1 AQUIFER COMMITTEE FUTURE 2 meeting had been made to disband the Aquifer 3 Committee, I think that should be reconsidered. 4 Let it go to inactive status until it needs to be 5 active again. 6 MR. DYSART: As I recall, I may not be 7 correct, but as I recall you had suggested since 8 the work was completed you didn't need to be on 9 here. It was not, as I recall, a decision of the 10 body to disband. 11 MR. SCHUBERTH: I think you're right. 12 MR. DYSART: What my sense is you want to keep 13 the committee sort of on spinning reserve so that 14 when there are matters that need its attention, its 15 available and otherwise, and to keep the people, 16 the members, informed of the progress. Okay. Is 17 that acceptable to the body -- it appears to be and 18 we will operate in that way. Thank you very much. 19 MR. SCHUBERTH: I want to just end this 20 particular discussion by thanking Doug for making 21 the very clear and focused presentation regarding 22 this issue. 23 I think it's important for everyone to 24 understand that there was no criticism leveled at 25 the original report. It is just that additional 76 1 PEER REVIEW 2 information from other sectors had come forward 3 that were thought to be valid and important. Those 4 suggestions have been integrated into a subsequent 5 study to be carried out, and that was actually the 6 ultimate objective of raising the questions, 7 whether that study was sufficiently comprehensive 8 to address everything that seemed to be known about 9 the aquifer, which isn't all that much. 10 MR. DYSART: Thank you. Next item, peer 11 review. Teri. 12 MS. LEFFEK: Well, Operating Guidelines 13 Committee was tasked to look at this. We made some 14 progress but not much. We are still waiting for 15 comments. 16 I'm still waiting for comments from all the 17 committee members. I did send out an e-mail and 18 posted on the website, under the Operating 19 Guidelines, a chart that Doug very graciously 20 provided us looking at different studies, what the 21 reviews are. 22 I have their comments on each of those to 23 suggest why what the Corps is proposing is not 24 acceptable, in their opinion, but not just to level 25 criticism, but also to make helpful suggestions how 77 1 REPORT UPDATES 2 to change that, so that what is the out there is 3 more palatable from the viewpoints. 4 So far we have very few comments. I'm still 5 waiting for further comments. That issue has not 6 been resolved. 7 MR. DYSART: Thank you. Comments on this? 8 Okay. Thank you. We'll, obviously, be looking 9 forward to more information back from your group. 10 Now we're ready for committee reports. First of 11 all, Bill Farmer, Beach Erosion. 12 MR. FARMER: The Beach Erosion Committee has 13 not met since the last time the SEG has met. We're 14 in the same stage that the Aquifer Committee just 15 recently came into, basically, we have recommended 16 the studies that we thought needed to be done, and 17 they're underway with the Corps and GPA. 18 Our next function, basically, is to comment on 19 the acceptability or otherwise of the final reports 20 that come out of all that's recommended. So we are 21 still awaiting the peer review process on the basic 22 Beach Erosion study, so we have nothing, basically, 23 to report as far as progress in that line. 24 MR. DYSART: Thank you, Bill. Anything on the 25 Communications. Dredging and disposal. 78 1 REPORT UPDATES 2 MR. BEASON: We're still waiting on the model 3 that's scheduled to come out late this year. When 4 that does, we'll have something to report. 5 MR. DYSART: Okay. Economics Working Group. 6 MS. JENNINGS: Thank you, Ben. We have a 7 meeting this afternoon at 2:00 o'clock. I'd love 8 for everybody to come. Bill Bailey, with the 9 Corps, will lead a discussion about how the Corps 10 plans to value economic -- I'm sorry, how the Corps 11 plans to value natural resources in the Tier II 12 EIS. 13 It's at 2:00 o'clock at GPA, and you'll just 14 go to the administration building and sign in at 15 the desk and wait on Cathy. She'll figure out -- 16 she has a place, I mean, a meeting room for us. I 17 really do hope folks will come, because we've 18 talked quite a bit about looking at how this can be 19 done, and I don't mean -- Bill, would like to add 20 anything? 21 MR. BAILEY: You did pretty well. 22 MS. JENNINGS: Thank you. 23 MR. DYSART: Is that it, Judy? 24 MS. JENNINGS: Except to encourage attendance. 25 MR. DYSART: How about Fisheries and Aquatic 79 1 REPORT UPDATES 2 Resources? 3 MR. BERSON: Very short report. We have not 4 met since the last SEG meeting. I hope, actually, 5 that unless the SEG decides to meet next month, 6 that we will have a meeting prior to the next 7 meeting. I'm going to throw that out to the 8 members. I'll be corralling you to get a date 9 after the meeting. 10 MR. DYSART: I've had several requests, Will, 11 wanting to know what was the song in your heart 12 this morning. 13 MR. BERSON: Oh, Nothing but Blue Skies. 14 MR. DYSART: Okay. That's a good omen. MTRG. 15 No Bo. Operating Guidelines, Teri. 16 MS. LEFFEK: Well, the chair has been 17 negligent in that I need to get approved minutes. 18 I don't have approved minutes in order to go into 19 detail about what we talked about. 20 We still have the peer review, which I'm going 21 to try to address that before the next SEG meeting, 22 unless the committee members want to throw anything 23 else out on the floor. 24 MR. REES: Speak up a little, Teri. 25 MS. LEFFEK: I basically said I don't have my 80 1 REPORT UPDATES 2 minutes approved. I don't really want to discuss 3 that in detail without everyone having approved, 4 unless the committee members have anything, 5 specifically, they would like to add. I don't have 6 anything further. 7 MR. DYSART: Let me just ask, is there -- 8 would it be appropriate for you to talk about the 9 -- can you give indications of what was discussed, 10 or the nature of the discussion, or would a more 11 appropriate situation require that you not comment 12 until -- 13 MS. LEFFEK: I think a comment, I didn't want 14 to say anything to give the impression it was 15 official, not having run it by committee members. 16 We talked about peer review, as I mentioned 17 earlier, did not make very much progress on that. 18 I hope that we are able to finalize that, resolve 19 that issue at some point. 20 We talked about the mitigation discussion a 21 little bit, the original concept behind that, 22 possibly recommending retiring that or changing 23 that. We also talked about a committee a little 24 bit as well. That's in a nutshell what was 25 discussed. 81 1 NEW BUSINESS 2 MR. DYSART: Okay. Striped Bass or Striped 3 Bass. Moving on under new business, the extent and 4 possible causes of marsh die-off, update on that. 5 Anything -- anybody anything to report on that? 6 Okay. The next item was to assess the future 7 need for the Communications Committee, possible 8 work for it to do to support the SEG's mission. 9 Chris and David Kyler suggested that. 10 MR. KYLER: I can't remember where this 11 happened here, whether it was here or the Operating 12 Guidelines meetings. It seemed to me reasonable, I 13 think I got agreement with Operating Guidelines, to 14 simply integrate that with the Operating Guidelines 15 Committee function. 16 As far as actual assignments for 17 communications, that's a different matter. We 18 feel, I think the consensus was, there's no point 19 of having a separate Communications Committee, 20 since it's been so long since they met. To the 21 extent communications issue come up, we felt it 22 would be best handled by the Operating Guidelines 23 Committee. 24 MR. DYSART: Chris. 25 MR. SCHUBERTH: I agree. If I remember right, 82 1 NEW BUSINESS 2 I just at the last meeting raised the question that 3 we've got this Communications Committee with Ben 4 Brewton as the chair. It met the last time January 5 7th, '00. Here we are in the middle of '03, and 6 then it was suggested, somewhere along the line, it 7 be folded into the Operating Guidelines Committee. 8 MR. DYSART: Any reaction, any thoughts on 9 that, any objections? Teri. 10 MS. LEFFEK: I think I needed to talk to Ben. 11 MR. SCHUBERTH: I can't remember the details. 12 MS. LEFFEK: The SEG can do whatever it wants. 13 I think somebody, I don't mind being that person, 14 should talk to Ben before we disband the committee. 15 MR. SCHUBERTH: I'm not sure we're disbanding 16 the committee. We're reconfiguring it into a 17 different context. With that reconfiguring it into 18 a different context, I guess Ben should be aware 19 this is happening. Put it to him as almost a fait 20 accompli, I don't think we're going to get a 21 reaction because he hasn't been here in three 22 years. 23 MR. BROWNING: I think the reason we talked 24 about it, Chris, was that almost everybody that's 25 on the Communication Committee is part of the 83 1 NEW BUSINESS 2 Operating Guidelines Committee. 3 MR. SCHUBERTH: Right, right. 4 MR. DYSART: Thank you, Donny. Let's explore, 5 do you simply want to communicate the feeling of 6 the body or do you want -- are you asking or 7 recommending that Teri discuss this with Ben 8 Brewton, or what is the will of this body? 9 MR. SCHUBERTH: My suggestion is that the SEG 10 go along with this suggestion with reconfiguring 11 the Communications Committee, in the context we 12 spoke about it. Teri is simply give Ben a courtesy 13 call to say this is what the SEG has recommended 14 to do, and we appreciate his past service to the 15 cause. 16 MR. DYSART: How does that sound -- nicely 17 stated. Any objections? If not, I will declare 18 that a broad consensus has been reached on this 19 issue. Teri, pass that information along. 20 Okay. How about the next meeting date? And 21 also please be thinking, it's been mentioned that 22 some studies would be about ripe for presentation 23 soon. Let me know now or by e-mail promptly what 24 might be ready for the next meeting. 25 August the 5th has been mentioned. The man 84 1 NEW BUSINESS 2 with a song in his heart concurs. Okay. Consensus 3 on August 5th. In addition to letting me know 4 other items, briefings desired and so forth, help 5 me if there's a strong feeling that there are items 6 that are kind of of a continuing nature on the 7 agenda that you want to reconfigure or slide off 8 the agenda. Let me know. This is your agenda. 9 Anything else to come before the group? Morgan. 10 MR. REES: I just request of Judy, we were 11 planning to meeting at 2:00 o'clock for the 12 Economic Working Group. I think that was probably 13 on the assumption we would be here until noon. 14 MS. JENNINGS: Right. 15 MR. REES: Do we want to meet earlier than 16 that? 17 MS. JENNINGS: Great suggestion, love to 18 compress time. 19 MR. REES: Anybody here planning to attend the 20 Economic Working Group who couldn't make it at 1:00 21 o'clock -- would that cause a problem with the 22 facility? 23 MS. VAUGHN: We had a meeting earlier that 24 should be cleared out by 1:00. It should be -- I 25 can't promise they cleared by 1:00. 85 1 NEW BUSINESS 2 MS. MOORER: Cathy, if it's not, maybe the 3 engineering conference room. 4 MS. VAUGHN: Depending on how many people. 5 MR. DYSART: Okay. The Economic Working Group 6 meeting is rescheduled to 1:00 o'clock today with 7 meeting next on August the 5th. If you'll let me 8 know what you would like on the agenda in the way 9 of briefings or something else. 10 MR. REES: Whispering in my ear -- why don't 11 we do it now? We have this room until 1:00. 12 MR. DYSART: I declare this meeting concluded. 13 You can without delay delve into the next meeting 14 if you desire. 15 16 17 18 19 20 21 22 23 24 25 86 1 2 3 4 5 C E R T I F I C A T E 6 G E O R G I A : 7 CHATHAM COUNTY: 8 I hereby certify that the foregoing 9 transcript was taken down, as stated in the 10 caption, and the questions and answers thereto 11 were reduced to typewriting under my direction; 12 that the foregoing Pages 1 through 85 represent 13 a true and correct transcript of the evidence 14 given upon said hearing, and I further certify 15 that I am not of kin or counsel to the parties 16 in the case; am not in the regular employ of 17 counsel for any of said parties; nor am I in 18 anywise interested in the result of said case. 19 This, the 22nd day of June, 2003. 20 21 22 ________________________ 23 Kathleen Dore, Certified Court Reporter, B-2041 24 25