Saturday, September 1, 2012

Finally, Into the Deep

Summer Research

Wait, I'm Going to Be Working on Physical WHAT?

The past two projects I wrote about were based on my Environmental Engineering coursework. Their primary purpose was to teach me something about this broad field, and as such they were not particularly applicable to anything outside of the classroom or lab environment.

In fact, it was not until the summer after my junior year that I got to work on an actual research project in a discipline related to Env. Eng. Though I had done research at MIT previously, these experiences had not really been in my field of interest (or, rather, had made me uninterested in their field): my freshman/sophomore year I worked in an environmental genomics lab, and my junior year I worked with health systems analysis. The lab environments and skills I gained from these experiences were invaluable, but I could not see myself in either profession long term.

In fact, I was waffling about where I saw myself in the wide spectrum of environmental engineering throughout most of my college career.

Until I was accepted into a Research Experience for Undergraduates (REU) program my junior year to work with a Physical Oceanographer, and my research focus swung a hard right to center on fluid dynamics.

The Project

I applied to this REU mainly because it was something to do over the summer away from college. By my junior year I was relatively sick of the campus; I had been there almost nonstop for two and a half years, only leaving about two weeks each year to go home. I wanted to get away, and I wanted a program that would make my trip worthwhile.

I was fortunate enough to be accepted, even though I was a professed engineer and the program was technically geared towards marine scientists. As a result, I was one of two students paired with a physical oceanographer (the other came from a physics background).

At first, of course, I had no idea what physical oceanography was or how my training could have prepared me for that field. I was vaguely confident that something I had learned in the past three years would come in handy, but I had no idea what. It depended heavily on the project I was to work on.

I was also deathly afraid of the prospect of fieldwork in a marine biology/oceanography lab, but that is a notion best left alone for now.

When I arrived at the program, I learned that I would be working on the Amazon River Plume. I also learned that physical oceanographers spend a good deal of time in front of their computers, and what my project would really test was my ability to do math and utilize MATLAB for days on end without driving myself crazy. As the only engineer in a program populated with scientists, it was nice to have something familiar to fall back on.

My supervisor had collected data from the plume the summer previous on an extensive research cruise. I was tasked with taking this data and "making something of it" (a common theme in my research since, it seems). I ended up flailing around with MATLAB for the first week or so until my supervisor and I decided on a more specific direction.

The meat of my project became looking at CTD (salinity, temperature, depth) and ADCP (velocity) data from the cruise to determine a method for finding the age and volume of the plume based on remote sensing data (satellite data). This involved a great deal of correlation analyses and verification of the cruise data against remote sensing data gleaned from various online sources. It was also a project idea to which I feel I had a significant contribution.

I have to say that the measurements from this research cruise were impressive. I learned in my first week the benefits of having an extensive data set--and I also learned that no research cruise can possibly collect as much data as a scientist would like for the analysis stage. Thus sometime in my early days as an REU student I began exploring the uses of remote sensing data.

Satellite data was an attractive option because it provided a significant amount of spacial coverage of any region of the world, really. There are enough research satellites in orbit now to give a good amount of information on, for instance, sea surface temperature. The trick is that they can only supply information about surface properties (and, you know, there are also these things called clouds which throw everything off now and again).

Given the amount of potential spacial resolution in comparison to the relative dearth of in situ data, I was very much attracted to the idea of figuring out how to extract more information from these research satellites. My project, therefore, became an attempt to link the data from the Amazon River Plume research cruise to remote sensing data in hopes of finding a correlation that would remove the need for the cruise data altogether.

The Result

Three months of data crunching, correlation testing, and generally plotting as many pretty pictures as possible resulted in a moderately robust relationship between K490, the light attenuation coefficient at 490 nm, and plume thickness.

You might be thinking "wait, where did that come from? Weren't we talking about plume age and volume ten seconds ago?"

Well, yes. But to get plume volume you need surface area and thickness, right? Satellite imagery can easily supply the former; my job boiled down to figuring out how to get the latter. This happened in several steps.

  1. Find a relationship between plume thickness (PT) and sea surface salinity (SSS) using the in situ data from the research cruise.
    • This was a reasonable correlation to attempt to find. You would expect SSS to be greater the more the plume mixed with surrounding waters, which would ideally occur concurrently with dissipating thickness.
    • It turns out that this correlation exists, but not as neatly as I would have liked. PT corresponds to a range of SSS--basically, the best you can get is an approximation of PT given SSS (to within 3 or so meters).
  2. Expand that relationship to one between plume thickness and colored dissolved organic matter (CDOM).
    • This step was actually a correlation between SSS and CDOM, again based on in situ data. This correlation was an incredibly nice exponential decay.
    • Explains a lot, since we expect a younger plume to have significantly more CDOM than an older plume given the fact that the plume is a nutrient, sediment, and organism-rich body of water infiltrating a relatively nutrient-poor environment.
  3. Finally, determine the relationship between CDOM and K490 as the final link in the chain of correlations from PT to K490.
    • At last, I was able to match the in situ data with remote sensing data. The relationship between CDOM and K490 was very strange--something logarithmic provided the best fit, which made little sense to me scientifically. I am still puzzling this one out.
    • Again, though, it makes sense that a relationship should exist since the optical properties of the water, represented by K490, are highly depended on what sorts of organisms and particles are in the water (CDOM).

In the end, therefore, I found what amounted to an equation to translate between PT and K490, an equation of debatable robustness. Had I more data from different seasons, and just more in situ data in general, this relationship could have been made more accurate. Still, no research projects is without its flaws, and the fact that a relationship of any robustness whatsoever existed was exciting to discover.

My exploration into plume age was slightly less complex, and generally involved applying ADCP data to satellite images of the plume extent. Once you have distance and velocity, calculating time is a simple multiplication step. The plume age I found, of course, was a broad estimate of how long ago a particular particle of water had left the mouth of the Amazon River.

The Aftermath


In short, there are several caveats to my results, but both are definitively results. They can be used by anyone interested in the plume's extent (for purposes of tracking organisms or particles/chemical reactions) to at least allow them an idea of the magnitude of the plumes impact on its surrounding environment.

This was the first real contribution I made to a research project, and I was able to create two posters to send to conferences based on my results from this summer alone. There was a real sense of accomplishment from unearthing these relationships and having a hand in designing the research I was conducting.

I was hooked on the idea of being able to do this again in the future. Even the prospect of sitting in an office for days on end doing data crunching was not enough to deter my interest in future work of this kind. I wanted to do useful work in the future, and this research project was the closest I had gotten to this goal.

However, I felt as if I was missing some of the more fun aspects of oceanography research; namely, the fieldwork and modeling. Thus I decided the summer after my junior year that I would return to research in the ocean, but only if I could fill in the gaps I had missed. I want to experience the system I study firsthand and get a feel for its physics before diving into the data crunching.

I also want to be able to model the system in order to discover more about it than fieldwork alone can reveal. I now have an idea of how to use cruise data and even satellite data to verify a model of any aquatic system.

With these goals in mind, I left the REU program completely satisfied, for the first time ever, with the outcome of a research project.

Saturday, August 18, 2012

They're Called "Instruments" Not "Machines"

Junior Year

Adventures in Environmental BioChemistry


My next major project as an engineering student began, coincidentally, one year after the week my sensor buoy group was formed.

The disparity between my junior year spring project and the buoy reflects the broad spectrum of expertise an Environmental Engineer is expected to grasp. The subject was environmental biochemistry, a class split between intense lectures (of the old-school-cold-calling style) and lab experiments (of the don't-let-these-incompetent-undergrads-die style). Both the lectures and the lab coalesced in the final three weeks as the students were instructed to find partners, design an experiment suitable to the class material, and implement it with whatever time remained in the semester.

I was incredibly lucky to be paired with a classmate who I was used to working with. We had done problem sets and smaller projects together because we took a similar approach to work that eschewed procrastination (the fact that the two of us lived in the same building was simply a perk). 

Neither of us was particularly interested in the biology portion of biochemistry; one lab class spent waiting for bacteria to grow was as much experience with microbes as we cared to get. So we decided to focus on the chemistry aspect, reasoning that it would take less time and generally be less of a hassle (no dependence on living organisms! Yay!).

This assumption turned out to be a little presumptuous.

Defining the Experiment


The first hurdle was getting an experimental design passed our professor. He was meant to be a "sponsor" (read: sanity check) of our final project. Being the opposite of procrastinators, we bombarded him with ideas as early as possible in the project. We brainstormed and conducted literature research for feasibility analysis, then presented the project topics to our chosen professor as soon as they took shape.

It was, oddly enough, a tug-of-war between feasibility and scientific interest. Given time and budget constraints, we needed to use the resources and environment close to our campus. However, we also wanted a reasonably engaging research question. Thus my partner and I discarded ideas we considered unfeasible; our sponsor nixed anything not scientifically intriguing.

It probably should have been the other way around, but our sponsor was rather excitable. And by junior year my partner and I were wary of taking on too much work for any given project (lessons learned from the sensor buoy, perhaps).

After numerous, long discussions with our sponsor, we finally settled on a project to test the concentration of DDT in the Charles River. You know, that chemical made famous by Rachel Carson in "Silent Spring." I could give you a picture of the chemical structure here, but I doubt it would be particularly enlightening. And if you're reading this, I am confident in your ability to utilize Google.

The main motivation behind the project was a combination of the lack of published measurements of DDT since the 1980s and various articles since then claiming that there was some DDT present in the Charles. Furthermore, during our literature search we came across reports on high DDT concentrations in the lower regions of the Mystic River, not too far north from Boston.

DDT, as an organic compound, is naturally hydrophobic. However, the possibility remained that DDT was concentrated in the sediment of the river with the potential to be suspended in the water column as the sediment was disturbed.

Simplistic diagram of the flux of DDT from sediment to river water. By testing both, we could get an understanding of its concentration in the water column and the correct direction of this flux.

It seemed reasonable to update the testing for DDT in the Charles River and settle fears of its presence once and for all. The key word there, of course, is "seemed." (Ah, foreshadowing).

Collecting Samples

It was early one Thursday morning in late April. Other groups were in the student lab, just beginning to set up their experiments and think about collecting data. Professors and one lab tech were on hand to prevent us undergrads from accidentally setting the lab on fire (or something).

 My partner and I, on the other hand, were out in a small motor boat on the Charles River dunking bottles into the surface water and a grab sampler into the sediment below.

It was our first experience at fieldwork, and and a relatively nice introduction. We had sterilized jars for sediments and old chemical bottles for surface water samples. We hijacked our sponsor to drive us and the bottles to the Sailing Pavilion on the river. We had already begged the Pavilion staff to take us out that morning, and to their credit they agreed. 

(Note: boathouse staffers are just about the best people ever. If you're a scientist or engineer working with water, they are your friends when it comes to sample collection).

The work was not too complicated; for various scientific reasons (reproducibility, breadth, flux calculations) we took several samples at three locations between two bridges in the river, steering clear of the bridges themselves. The samples were from both sediment and surface water, and each one had its own quirk. The sediment was, disturbingly enough, the consistency of toothpaste, completely black, and pungent (of the anoxic variety). The water was basically cold (I had the pleasure of holding the 6 L bottles underwater until they were full, and my hands were happily numb by the end).

Sample locations, taken before and after sampling. Notice anything strange about that last location (in green)?
After a couple hours of hard work, we got all of our samples out of the river and back to the lab. It was time to figure out what, if any, DDT they possessed.

Sediment samples were put in glass jars (left) and covered in foil to stop photosynthetic activity. Water samples were put in jugs (right) which were thankfully already light-proof.

The Chemistry

Our experimental design was slightly unusual at the suggestion of our sponsor. His lab had begun to use polyethelene strips to extract organic chemicals from field samples (the idea being that organic chemicals were more attracted to this material than the sample material, be it soil or water). It was a great way, in theory, to concentrate the chemicals onto a small strip of plastic which could then be worked on in a lab.

Of course, there was a lot of extrapolating and concentration conversions behind this method. But I won't go into those; they aren't that difficult to reproduce, and my lab notebook is somewhere else at the moment.

We put polyethelene strips in each sample and left them for about a week in a tumbler (a plastic container lined with foam, suspended between to two wood sheets and spinning slowly with the help of a motor; truly an invention of necessity). The timing was unfortunately short; we would have rather kept the samples and the strips together for longer to ensure their equilibration, but the end of the semester was coming fast and luxuries like time were fleeting dreams.

After a week, then, we removed the strips from our samples and extracted them with hexane. All of this went rather smoothly; labeling and cataloging samples is a rather relaxing task.

And then it came time to analyze the extracted chemicals.

Troubles with a GC/ECD

Our analytical method of choice was a massive, old Gas Chromatography Electron Capture Detector (GC/ECD) machine. I mean instrument. I mean infuriating fossil of an instrument.

In short, it did not work. At least, not well and not often. To my chagrin, I was not allowed in the lab to do sample analysis except when either a professor or lab technician were present. This meant that I was not allowed to work outside of normal work hours, because it was highly unlikely any professor would be awake at 2 am. 

Nor was I allowed to simply work with the GC/ECD until I figured it out. There was an understandable fear from the lab techs that I, as an undergraduate, would break the machine (er, instrument) if I played with it too much. Unfortunately, this meant that I was entirely dependent on the lab techs to troubleshoot the instrument when things went wrong.

Things went wrong a lot.

The basic idea was that we had prepared standards of known DDT concentrations (as well as its breakdown products DDE and DDD) to run on the machine as a calibration. Once we had these standards measured, we could run our samples and move on to analyzing the data. 

Elution times of our DDD, DDE, and DDT standards. We had to test a wide range of concentrations and isotopes of these compounds to ensure we were testing for all possible forms of DDT.
However, the GC/ECD refused to return anything reasonable for our standards. The problem appeared to be a combination of the type of coil it used to analyze our particular range of organic compounds, the cleanliness of its interior, and its impressive age. Unfortunately, we could only fix two of the three.

So the last couple of weeks of term saw me and my partner in lab as much as we could be, only breaking to attend other classes in any given day. We spent all of this time troubleshooting the machine with the lab technician--and getting lectured by him as to why we should refer to it as an "instrument". Apparently it is a sign of respect and makes your experiment more likely to succeed.

I'm not sure how much I believe this, given that even once we began calling the ECD an "instrument" it still stubbornly refused to work. Maybe "temper tantrum" would have been a better label.

The lab tech, however, was a wonderful human being who stayed with us in the lab from early morning to evening. And with his ministrations, the instrument slowly emerged from its tantrum and began giving us actual results.

The End

With maybe four days to go, we finally had acquired data on our samples. What we did with this data is not particularly important, as there wasn't time for much processing and what we did manage to collect was rather sparse. So I will simply state the results of the experiment and leave it at that.

We found, to no one's surprise, that DDT is not at high enough concentration in the Charles River for anyone to be concerned. Its breakdown products, DDD and DDE, are present at low concentrations. Again, however, their levels are not significant enough to worry any casual swimmer or fisher in the river. They were in rather unexpected locations, however. DDD, the anoxic breakdown product, was in the surface water at higher concentrations than DDE, the oxic breakdown product. And the reverse was true for the sediment.

I am still puzzling that one out in my head, but without samples from farther upstream there is no way of knowing for sure why this counter-intuitive result exists.

And I've moved on to larger bodies of water since then, so I am not sure that I will ever return to the Charles to take the necessary measurements. The application of our experiment, after all, turned out to be rather slim. However, as an introduction to fieldwork and the tricky business of getting lab instruments to do as I please, this was a fine project.