Cape Cod Trip Large-scale Aquifer Sampling

all right you're probably wondering why there's so many pipes out here um back in um back in 1985 80 actually earlier in 93 84 we uh we had we had just we had discovered the sewage plume we were beginning to get a feeling for what it looked like and everything else and um one of the big questions at the time in in the hydro groundwater hydrology was the rate at which contaminants disperse you know how how quickly will if you put in a if you had a spill from a septic tank or a you know fuel spill or whatever else and and it got in the aquifer and it begins to migrate with the natural groundwater flow how quickly would it get diluted and dispersed to the point where it wouldn't be a problem and there were a lot of ideas about how quickly that happened and what caused it and what they what a number of folks were able to demonstrate is that the primary cause of the dispersion in groundwater unlike in surface waters or in the atmosphere which is turbulence there's no turbulence in groundwater it moves through slowly you probably have you know heard about laminar flow it's really kind of almost limiter is is the actual heterogeneity of the subsurface in other words even the most uniform quote unquote aquifer like a sand and gravel outwash like this is actually has a fair amount of lensing and layering because it's a natural geologic deposit you know it has beds you know sedimentation beds from running water so some some pioneers like lynn gelhar i don't know if lynn gelhardt comes into the lab at all or not anybody know lynn gelhart professor at mit he's retired now was one of the people who actually was able to demonstrate how you could predict the rate of dispersion if you could characterize the heterogeneity of the aquifer so if you could find some way of statistically characterizing how the hydraulic conductivity varied from place to place and how long the lenses and layers were and how thick they were that you could actually predict how rapidly contaminants would disperse so he and a few others had come up with various stochastic uh models of that process and the idea was that you know is there any reality to them at all so um we teamed up with uh professor gelhart at mit one of his phd students lynn uh steve garrobinian and i and a few others uh that we hired to help us out and we ran a tracer experiment to try to measure the rate of dispersion in the aquifer so our idea was we would create our own little plume and we would monitor it as it moved through the aquifer and actually by collecting water samples in three dimensions so we'd have a detailed 3d snapshot of this tracer plume we could actually measure how rapidly it dispersed and spread and then we could compare that to other measurements of the aquifer's heterogeneity and see if the theories were right so it was it was a a massive tracer experiment so what we did is uh we picked this spot somebody had asked me why did you pick this spot uh well we picked this spot to run the tracer experiment for a couple of reasons first of all it was an abandoned gravel pit and uh so it was flat and easy to work in secondly it was on state-owned land so we had permission to do it and thirdly uh the groundwater was already contaminated here because of the sewage disposal so we were able to get permission from the state to create our own plume within a plume you know if we had gone through a pristine aquifer they would have gone are you nuts you want to add a contaminant you know so uh so we were able to do it right here so what we did on uh in july 1985 we uh see the three wells with the yellow labels up there the three wells those are all screened uh just below the water table they're shallow wells we uh we pumped two thousand gallons of water out of the aquifer into tanks this was clean groundwater pristine groundwater we added a number of traces to the water and in particular the one i'll talk about today is bromide uh like chloride it's very non-reactive so it's a good tracer of the water and bromide levels in the aquifer naturally are very low so we had a good signature of the water we put in as opposed to the background water right and then we uh we injected it over a very short period of time about a day took about about 19 hours into those three wells so what we did is we created a pulse you know on the time scale of years essentially instantaneous pulse in the aquifer a bromide cloud and if you just did some simple calculations it would have been about four meters by four meters by three meters now it didn't go in that thing more like a beautiful box but that was roughly the size we expected it to be all right and then what we did is we then let the natural flow carry it downstream how did we know which way it would go we'd done all those water table wells first we had mapped the water table out here in detail and we could predict which direction it would take more or less all right and uh so then we started to track it and the way we did that was by using these multi-level samplers that we had drilled out ahead of it right so you can just imagine this process we um we inject the tracer and we had put in about 10 rows of samplers and we would come out here and look at them wow look at all those wells isn't that cool there were about 80 of them at the time and we injected the tracers and then right away i think the first time we did a sampling was 13 days later we went out and sampled as many of these multi-level samples as we could ran back to the lab and then analyzed the samples for bromide and tried to figure out where this tracer cloud was how big was it and on that basis we then would build more samplers get out ahead of where the cloud was going to go next based on our predictions drilled more samplers and then we leapfrogged for two years that way all the way down across the gravel pit so all these samplers were put in as part of that treasure test back in 1985. um these are all multi-level samplers so you're looking at a well field here this we've now since run about 70 tracer experiments out here so we've added some as we go along there's about a thousand out here which means there's 15 000 sampling points in the gravel pad that's a good question i i would guess uh it's hard it's hard to really put a number on it but let's just say a million dollars i mean i just i have no way of really telling you because it's old dollars versus new dollars and you know it's really hard to say and then we had a lot of researchers work with us that i didn't control their money you know i didn't fund them so uh my budgets were always less than a half a million they were they were uh typically a couple hundred thousand but then that was 1984 dollars you know so i don't know what it would be today so did you prove the stochastic model i came out uh frequently closed the one thing that surprised us was that the heterogeneity was much even though it looks heterogeneous i was going to take a pit for you today and i said no i'm not taking a fit for them today but uh it looks very very heterogeneous when you look at the lenses and layers but it's not that i'm a hydrological point of view it's pretty uniform and then we ran another test in 95 93 to 95 another large-scale test where we did different reactive species but we again did bromide and got a very similar number uh so and it matched up again the stochastics pretty well so in fact the number of mit i think i want to tell you three mit phd thesis arose out of this test um so uh it was a fair number of mit folks were involved uh mainly lynn gelhardt students um and um so the way we sample them you can imagine trying to sample this the way we sample them you can imagine we come out here and we gotta take a snapshot of where the cloud is and we had a team of five of us we had two carts like this and instead of something one tube at a time we would hook up the milking machine like this and we had a chart that told us how much water we had to purge out to get rid of the old water and we would turn on the milking machine and they would all start to pump if we're lucky okay and um so the person would sit here they were they were the they would attach the milky machine and run the cart in order not to pump too much out of any one port because some yield really well you had a little cliff where you could stop the flow on one that was getting ahead of the others okay and we had these 60 milliliter bottles here what we would do is we would shove these in here and these were our rinse racks so let's say the chart said at this distance with this depth we need to pump out about 300 milliliters before you can take a sample we knew that was say five wrenches so we would shove this in we would fill them up when one side of the fill up would close it off when we had it all filled we would do this dump it off it would go into a bucket you shove it back in until you did your rinses then the person behind who was the bottle capper and the that's their job the cap bottles for four days straight um would have loaded a rack with the sample bottles in you'd shove that in fill your bottles up give it to them they'd cap them all like a madman while you went on to the next one and we would do that uh and we had uh so we had two carts and then we had the field general um anybody live at mccormick no that's an undergraduate dorm i guess right you know kathy hess yeah yeah kathy hess is the house mother at mccormick right or whenever you guys dorm house master she's my house she's my co-worker i call it val's mother uh there are some times when we sampled 300 multi-level samplers three samples each port in three days so we could really fly take a snapshot we did that i think 18 times in the course of two years very quickly because we didn't have much time we had to get it going right away we also went to the university of waterloo who had done some similar experiments and um and you know really borrowed a lot of their ideas to be to be honest so uh you know we won't go into it in any more detail in that just to give you a sense for how it worked on page 10 it just shows you and 11 just shows you a couple of the results the uh and i won't go any further than that except to say that what we've discovered was when you injected the tracer as i mentioned before it was a pretty small volume within 30 days the tracer cloud was already pretty large in fact uh why don't we get uh we might as well do this it won't take a second see if i can figure this out here hey pete can i send you up as a runner and eric i want you to find i want you to go up to uh row one up there at the beginning and i want eric to find row 17. he'll say 17 on the covers faster and faster think of them as little bromides okay so there you go so by by 33 days the tracer cloud had moved from those injection points to where the one milligram per liter contour would have been this far south and that far north so it spread very very rapidly you know in a short period of time so by 33 days this claw was already like 100 feet long but very very narrow it never got very narrow and then they migrated forward at a rate of one and a half feet per day so uh pete go to row 20 and eric you got to go to row 37. i think i have to leapfrog so by 139 days row 27 actually it's going to be beyond all of us here i think to row what did i tell you no you're 20. and he's 37. so you can see how how quickly it it uh spread in the longitudinal direction that's exactly what the stochastic theory said that most of the spreading should take place in the direction of flow and the plume actually never got very wide horizontally and in the vertical it stayed very very narrow uh it was only about uh six or seven meters thick there he is so that's how that's how long on the 139th day our sampling had to encompass all the samples from beyond pete to beyond eric and that was only 139 days okay thank you we always did it with five people we always did it with five people wow no we just we just worked an extra day yeah we just worked faster we got pretty good at it we were damn good at it we collected 100 000 water samples for the experiment