CHUCK NEWELL: OK, well let's continue with the way to understand our world of sources and plumes and attenuation, but maybe with a different model, a different way of learning. DAVE ADAMSON: I think you're talking about big data here, right? CHUCK NEWELL: That's right. You sort of think about our domain as starting with that 40 mil VOA vial that gives us our concentration, right, and we use this to look at temporal trends and understand the spatial distributions of our plumes. In some ways, it's a little bit like the human body, where we're starting with cells, then organs, and then, well, us. DAVE ADAMSON: Is there some sort of connection you're trying to make here? CHUCK NEWELL: Well it's maybe that the medical field is very big data intensive. If you want to know a prevalence of some disease or if they want to know if a treatment will work, they go to epidemiology studies. You know, they crunch the numbers. And in the groundwater field, we've been doing this for a while, but recently, there's been a dramatic increase in our big data-ness factor, OK? DAVE ADAMSON: Big data-ness? So another made up word from Chuck-- sounds interesting though. So I'm guessing you might be shifting to paradigms or something like that next? CHUCK NEWELL: OK, you're really prescient. You can really see in the future. So let's go look at this really good book called The Fourth Paradigm of Science, and it's the idea that science has sort of gone through these different waves and so they defined the first paradigm as what? DAVE ADAMSON: Experimental science, so running experiments. Think about like Mendel in his lab with beans, doing genetic experiments. CHUCK NEWELL: Pretty good. And now the second paradigm, theoretical science and the idea there, it's Einstein smoking his pipe and he's writing these equations on his blackboard and thinks and then he writes E equals mc squared and he's done. DAVE ADAMSON: Computational modeling, that's the third one in here. So where you got all this power in these computers, you can set up these really complex models to describe stuff like the weather, climate models, things like that. CHUCK NEWELL: So then the book goes in and says, we're on the cusp or we're starting to do this fourth paradigm of science and it's these application of these massive database that you can mine, and you can learn, and you can understand. Like one of the examples in some of these books was that by mining big data, that Walmart knew if a hurricane was in the Gulf of Mexico, people wanted to buy strawberry Pop-Tarts and so they would stock those in there, getting ready for the hurricane. So we can learn some of the same things except for plumes and things like that. So let's go back in time. We've been doing some of this for a while and they've been very important studies for us, but this is a particular study done by the Lawrence Livermore Lab. David Rice was the author and he looked at a couple of hundred of these benzene BTEX plumes. How long are they, is what they asked. So this was a drawing that we did for the American Petroleum Institute, but looking at this, we're looking at different buckets or different lengths of the plumes. So what's the key point here? DAVE ADAMSON: Well, I mean if we're looking at plume lengths there on the left, it looks like most of them are sort of in the under 200 foot range. CHUCK NEWELL: And this was an amazing revelation. These plumes aren't really long, but what was perhaps even more important was these temporal trends. They took all their sites where they had some data and they put them in these four buckets. Is it expanding, stable, shrinking, or exhausted? Dave, what was the big news out of this? DAVE ADAMSON: Only eight of these, in this case, were in the expanding category. CHUCK NEWELL: So some good learning just from sort of mining those databases and doing this fourth paradigm of science. Now they've also did some plume-athon studies back around the year 2000. This is, again, from the Lawrence Livermore guys. Walt McNab's the key author. But looking at chlorinated solvent plumes and they're seeing how long are they. Let's look at this median line, in terms of the median length for these-- 10 part per billion plumes. What was that medium length? DAVE ADAMSON: About 1600 feet and it decreases a little bit as you expand. After 1,000 ppb, it might be 650 feet, so pretty long plumes. CHUCK NEWELL: And just much longer than the BTEX plumes, right? So that was in there. They were also interested in this question, is we do the daughter products to get way ahead of these plumes. So they looked at that and what were their conclusions here? DAVE ADAMSON: Well here it's sort of interesting that the parent plume in these cases was the longer one in the majority of them, 73%. Only 27% was that daughter plume actually out beyond the parent plume CHUCK NEWELL: Yeah. So gives some insights. Maybe we don't know the exact processes, but this is what was out there and so an important learning. Let's go to another study. This is concentration versus time. This was a study that I did. Dave, what's on the graph on the left? DAVE ADAMSON: We got normalized concentration for a bunch of different wells so each of those red lines is sort of an individual well and you're plotting that concentration over time. So as concentration goes down, it might-- it gets lower. It's going to fall below that blue line. As it gets higher, falls above the blue line. CHUCK NEWELL: OK and a lot of scatter on that data. If you really want to know the trends, what do you do? DAVE ADAMSON: You ask the man. Do the Mann-Kendall trend analysis that we talked about it in another lecture. CHUCK NEWELL: So we did this, and if you look at the table there, let's look at all of these plumes. Of the 23, only three of them had expanding plumes. And the rest, there are two no trend, but most of them were stable, probably decreasing, or decreasing itself. So you can look at these things. So that was an important application of data mining and seeing what was done. Now let's move on to sort of more recent times and the emergence of this really amazing database called GeoTracker. Every groundwater sample that's been collected in California since about 2001 gets into this database. There are a total of 60,000 sites out there. There's 12,000 of them with this electronic data, is that right? And then this was used for this paper, "Progress in Remediation of Groundwater at Petroleum Sites in California," by our colleague, Tom McHugh. But each dot there is a site where they did the sampling in GeoTracker. And so we can-- want to blow this up a little bit more, get a sense of what we're looking at? So we're to go by-- each month is a sample that's taken and so you can go through here and just sort of see these things move. Each one of those is the field crews are going out there. They're collecting those samples, or sending them to the lab and then that data makes its way into this big database that you can do. And that's what Tom McHugh did to sort of understand this. Of course there's some things there. You can get a sense of what's going on. Let's look at this slide here. It looks like that there's this dot out in the ocean. DAVE ADAMSON: Those guys did some extra effort to go get that data point, I guess. CHUCK NEWELL: I guess. I have in my mind, the field guys go to the boss. He gives them those GPS coordintes and they say, but boss. He says, just do it. So they rent a boat and they go out there and they do their sample and get the slug test, high permeability. What do you think? DAVE ADAMSON: Yeah, yeah. Probably. CHUCK NEWELL: OK. So that's GeoTracker and so then Tom McHugh took this data and he said sort of what is the typical median concentration of benzene in California? How did it change from 2001 to 2010? DAVE ADAMSON: The expectation, it's going down over time, right? CHUCK NEWELL: Let's see if that really happened. What does the big data tell you? So here's this graph, median of the maximum concentrations at all these different sites. About a 10 year time interval on the bottom, and yeah, it goes down a lot. It's going from about 2000 to, what do you think, about 200 micrograms per liter? DAVE ADAMSON: Yeah, something like order of magnitude. So each individual point is the representative concentration for that sort of 1,128 benzene sites. CHUCK NEWELL: Yeah. And this was-- the same sites were used, so this analysis, they kicked out sites that were closed halfway through or new sites that came in. But this is what these trends say at these sites that were open. Now let's look at MTBE. You think you'll see the same thing? DAVE ADAMSON: Well, yeah, MTBE was starting to get phased out, so it should decrease. CHUCK NEWELL: Looks like it's even a bigger increase, maybe 2,500 micrograms per liter to something that's well below-- what do you think-- 50 or 25? DAVE ADAMSON: Yeah, yeah. So a couple orders of magnitude in this case over that relatively short, short period of time. CHUCK NEWELL: And then he was able to go in there and look at the degradation product from aerobic degradation MTBE. And so what did he see with this? There was an increase and then a decrease? DAVE ADAMSON: Yeah, so I guess you could expect to see some TBA formed while MTBE was being degraded, but then eventually seeing a decrease in that as well. CHUCK NEWELL: So some remarkable stuff that he finally actually put a model to this, right? And he says, I can fit this curve by turning these dials, which is the two inputs or what is the half life of MTBE, and that's creating, basically, this TBA. And then how quickly does the TBA itself go away? What are the decay coefficents? DAVE ADAMSON: Yeah, we about a two year half life for MTBE and 3.5 years for TBA. CHUCK NEWELL: Yeah. so some really amazing stuff that's coming out, looking at this big data from GeoTracker. Now you actually use GeoTracker for a different question, 1,4-dioxane. DAVE ADAMSON: Yeah, a lot of people interested in 1,4-dioxane relative to chlorinated solvents. There's a lot of co-occurrence of those at sites and there's a big worry that the dioxane plume, because of the way it could potentially migrate, would be way out ahead of the chlorinated solvents. So what we did was go into GeoTracker, tried to take what the data really showed to us. And it's pretty interesting stuff because if this is a graphic from a paper that we publish, really, only 21% of the sites fell into that sort of common conceptual model, where the dioxane plume was longer. 17% of sites, they were about the same length and what is really interesting then is 62% of the sites, the majority of them, were actually able to say that the chlorinated solvent plume itself was longer than the dioxane plume. So sort of positive news in terms of dioxane. CHUCK NEWELL: Now part of that may have become the dioxane may have been newer than some of these solvents or used at a later date, but also you went to GeoTracker and the GeoTracker data said, hey, this stuff's attenuating, right? DAVE ADAMSON: Yeah, and we have a lot of interesting information on attenuation, but one of the things that we're also able to do is then say, what's important for promoting attenuation? And so some of this data, we also got from Hunter Anderson at the Air Force and he did some great analyses with these, and that's what's shown on here. But the basic message is that dioxane attenuation was positively correlated with the presence of dissolved oxygen, so promoting that aerobic biodegradation, and negatively correlated with the presence of metals and CVOCs. So good indication that the field data is sort of matching up with all those great lab studies that we've seen at various research institutions and so forth. CHUCK NEWELL: OK, and what metals are you talking about here? DAVE ADAMSON: We're talking about things like chromium and things like that. CHUCK NEWELL: Very cool. OK, well let's wrap up these big data studies, give insights on how big plumes are and how fast they change and what they do. DAVE ADAMSON: And we've got numerous studies that are out there looking at a variety of different contaminants. CHUCK NEWELL: And finally this GeoTracker sort of a database, really an important resource for understanding how plumes work.
