Jonathan Clough is an environmental computer consultant based out of Warren, Vt.
Monday, 22 Jul 2002
The first thing you should know about a week in the life of an environmental modeler is that it is remarkably glamorous. Straight up glamour. Glamorous enough that 14-year-old girls write me fan mail and ask for help getting started in a career as a modeler.
Okay, I really should admit that this only happened to me once — last week, in fact. Given that the girl in question was actually trying to become a fashion model, she turns out to have been a bit misguided in her Internet quest. (Let’s hope that her future net searches turn up more productive, though equally benign, email partners.)
I also should admit that waking up and commuting across the hall to the little room with my computers does not fit many people’s definitions of glamorous. Perhaps “hermit-like” would be more appropriate. The severity of Monday morning is upon me so forgive me if I have stretched the truth a little bit. Given that the Red Sox dropped two 9-8 heart breakers in a row to the hated Yankees, a little poetic license to improve the author’s mood seems warranted.
Anyway, I spend my workdays making and applying computer models of environmental systems, primarily as a consultant to the U.S. Environmental Protection Agency. Most of my work for the past six years has been in producing models of water systems and the effects that various toxins have on the critters that live in the water. If that doesn’t inspire a young girl’s heart, I don’t know what would. (That may actually explain my social status in junior high, but that’s a completely different story.)
Generally, what our computer models try to do is to integrate specialized scientific studies into a larger framework so that difficult policy decisions can be made. Often times, public policy makers want to answer questions about complex environmental systems — complex enough that the answers aren’t necessarily intuitive. While scientists may have studied many unique aspects of the situation, these studies, taken in isolation, do not answer the question at hand. Environmental models provide a framework to integrate each of the relevant scientific studies into a single result that is useful to policy-makers. And that’s where I come in. Somebody’s got to be the computer geek on these projects.
Lately I have been spending most of my time modeling a fairly high-profile site in which a company dumped many tons of PCBs in or near a river from the 1930s to the 1970s. Some questions about the site can be answered without the help of modeling. Should the most polluted areas within the site be cleaned up? The answer is obviously yes. But when the question becomes, “How many miles downstream should the company clean?” the answer is not immediately clear — and hundreds of millions of dollars are suddenly at stake. This is where our environmental models come into play. These models try to discern what would happen under different cleanup scenarios and what would happen if those downstream miles were not cleaned up. Given the dollar amount on the line and the near certainty of litigation in this case, I’m afraid that I can’t really say too much more about that project here, so I’d better return to the travails of the present day.
Travails: a gray and foggy Monday morning, those discouraging Red Sox losses, a broken-down car, a mid-summer cold. Happily, the birds are singing outside my open windows, and they’re rapidly improving my mood — especially the hermit thrush. Did I mention that I live in rural Vermont? I have been working from home for the past six years, which enables me to choose where I live while keeping my job intact. This seems like a fairly unique benefit in the history of employment. The absence of a commute is certainly a wonderful timesaver. But there are drawbacks as well. I’ll get to those later. After all, I have diary entries to write all week.
And here, noble reader, is where you are in luck. This week of writing diary entries for Grist coincides with an intensive week of data mining! Being an environmental modeler requires an intimate relationship with all kinds of empirical data and this week I am tasked to have my hands elbow deep in the cow, so to speak. (Such colorful colloquialisms prove that Vermont living is starting to rub off on me.) Stay tuned for such exciting episodes as:
- Units mayhem!
- Where are the data quandaries?!
- Data entry glitches!
Actually, I promise it will be more interesting than that. Umm. I’ll try anyway. Stay tuned.
Tuesday mornings always find me feeling a bit behind the eight ball as far as work is concerned. It’s a problem of my own making: I attend a yoga class that starts at eight in the morning. What with socializing with classmates after the session is over, getting my mail at the post office, and constantly running into friends and acquaintances, I often don’t manage to get down to work until 10 in the morning. This means if I’m going to bill eight hours I’ve got to work until 6 p.m. straight without taking a lunch. Ouch!
Those darned acquaintances. You don’t realize it when you live in a city, but small-town living is all about running into folks wherever you go. This can be a drain on productivity, though certainly a pleasant one. One of those omnipresent Internet lists titled “You Know You’re From Vermont If” includes the following item:
It takes you three hours to go to the store for one item even when you’re in a rush because you have to stop and talk to everyone in town.
This is true.
But I manage to give my yoga practice such little time in my life as it is, that I just can’t give up Tuesday mornings. Even if I find myself starting a bit late, I generally have a better frame of mind to attack my daily tasks.
I started to practice yoga a little over 10 years ago, and at different times in my life I have managed to dedicate quite a few hours to it. From my first class, I have been fascinated by the effect the practice has on my mind. In a word (or four), it shuts it up. My own mind’s tendency is to be overly rational. That I gravitated toward a job in computer modeling is primary evidence of this. I’ve observed that a hyper-rational mind often comes with a certain degree of detachment from the world, and an unhealthy dose of egotism as well.
Through the practice of yoga, I’ve started to learn that when you manage to stop this constant blather from the mind, a pure layer of perception can start to come into your life. It is from this pure perception, unmodified by thought, that one starts to acutely feel the effects of one’s actions.
In my experience, such pure perception makes a yoga practitioner feel the effects of violence more acutely. I have heard several people discuss the fact that they feel vulnerable after finishing a yoga class. This is because they have been robbed of the shield that is their constantly moving mind. Without this shield, a practitioner can see every homeless person, smell every car fume, or feel the rage of each angry driver for what those things are.
People who start to perceive the effects of violence within their lives will usually try to eliminate it, generally starting with their own behavior. This is probably the most basic tenet of yoga ethics, ahimsa or non-violence.
This brings me back to my work. It is all fine and good to try to understand the consequences of various actions as they apply to complex systems. But to make this analysis matter, a set of ethical precepts must apply so that each consequence has meaning. And these ethical principles can become crystal clear when you manage to make your mind quiet down for a while.
That’s the way that I see it, anyway.
My two office-mates are a bunch of sad sacks today. These guys look like the Larsen cartoon about a boneless chicken farm in which the chickens lie around like big blobs. They are suffering from the fairly intense heat and the fact that I have no car to drive them to Blueberry Lake.
Still, they are a good kind of colleague to have. They don’t let me go a day without taking a walk. They never object to the music that I play. True, they do bark during conference calls occasionally, and I wish I could do something about their breath. But they are close to perfect work-mates.
Okay, now I’ve really got to get down to work.
Tuesday, 23 Jul 2002
Wednesday, 24 Jul 2002
If you were wondering, after yesterday’s report, I do have some non-canine colleagues; they’re spread out all over the country (and Canada). Over this past year, I have worked closely with a team comprised of folks in British Columbia, Washington, D.C., the other Washington, Mississippi, Tennessee, California, and several other U.S. states. It is an interesting setup, a truly virtual office, and it has both benefits and drawbacks.
Next week, many of us on the team are getting together for one of our periodic modeling meetings. Although these meetings can be a grind, I tend to look forward to them. Face-to-face time is certainly helpful when dealing with colleagues, and allows for that personal touch that is usually lost over email. Unless each email is composed very carefully, the format cannot reliably and accurately convey the author’s tone. Humor and irony can be dangerous techniques to use over email. I generally refuse to use “emoticons,” especially in my work emails, for while they are a safe way to imply a joke or irony, they may well leave the impression that the author is a bit goofy. :-). When I send emails to my colleagues I try to be sure that I’m not sounding angry or demanding (unless, of course, I am feeling angry or demanding).
Phone conferences tend to become a way of life for the home-office worker. These are certainly productive, and can help a virtual office come together. Our team has a weekly modeling call which allows each member of the team to communicate with the others in real time. Given the tendency of these meetings to drag on, though, a decent speakerphone is a must.
Today, I’m anxious to hit the ground running. There was a bit of spinning of my wheels yesterday as I tried to move forward on several fronts at once. This happens to everyone, but I think that it makes me feel a bit more anxious than many. If you work in an office, people at least know that you were there at your desk pounding away for eight hours. When you work out of your home, concrete progress on your tasks is the only thing that your clients can be sure of. Working from home is a little bit like being a student all the time. The upside is that if you need to take a break, you can generally do so at any time. The downside is, you constantly have deadlines hanging over you, whether it’s the weekend or not. What’s more, you are rewarded for the results of your effort and not for the effort itself.
If working from home is like being a student, most of my tasks are like a term paper or a thesis: large projects with a deadline that is usually several months away. When you consider that I may have four or five projects going at once, you can infer that the process is not exactly stress-free.
Another reason to get a jump on the day today is that I promised my office-mates a walk in the woods later on. In my case, that’s right out my back door.
Getting to choose your own location is definitely one of the most significant benefits of working from home. Until recently, I worked out of a home office in Boulder, Colo., a location that I loved. Housing costs and traffic jams finally drove me out of that region a couple of years ago, and my wife and I moved to central Vermont. It’s been difficult to adjust to the lack of sunny days here, but now that we have gotten used to it (mostly), we’re quite happy in our new location.
And we just have to walk through our back yard to bushwack up the little mountain called Warren Pinnacle. That is an important goal of mine for the day. The car repair is proving to be costly so I need to stretch my stressed-out legs.
Thursday, 25 Jul 2002
“Everybody loves an environmental model until it tells them to do something they don’t want to do.”
One of the models I have been working with for many years is currently undergoing a peer-review process; the quotation presented above has been the dominant response of one of the reviewers. She is right, of course. In the abstract, environmental models strike many people as desirable, because they integrate a wide array of scientific research into dynamic models of complex systems. Plus, they usually come with plenty of nifty graphics and colorful animated maps. As modeling techniques become ever more sophisticated, it feels as though we are making genuine progress in the realm of scientific inquiry and discovery.
However, as soon as an environmental model makes a prediction that leads to a consequence that is costly to industry or property owners, the model becomes quite unpopular indeed. In an effort to ward off the consequence, every underlying assumption behind the model becomes subject to deep and occasionally unfair scrutiny.
For example: When I was working for a consulting firm near Washington, D.C., I was sent to the U.S. EPA offices downtown to help respond to industry criticism of a particular modeling analysis. This analysis had led to an emissions regulation that would prove costly to certain industries. The industries had responded by creating six file-drawers full of reports and comments as to why the EPA’s analysis was bogus and why the regulation should not be implemented.
Looking through these file drawers, I was amazed at how every portion of the analysis was attacked, in many cases just to create additional pages of comments. Being new to the business at the time, my responses were limited to those comments that were completely off the mark — that is, based on a misunderstanding of the analysis or simply on errant science. There were many comments that could be dismissed on this basis. Still, the sheer quantity of these comments and the requirement of the agency to respond to them certainly managed to gum up the works and delay the passage of the regulation.
The model that was being attacked in this manner happened to be a fairly simple analysis. The problem can become even more acute when dealing with a more complex model. As one adds parameters to a model, more data and analyses are required to back up these parameters. While the added complexity may improve the model’s results, every additional level of sophistication provides more fodder for attack.
Of course, models can be attacked on the basis that they are too simple, as well. Look, for example, at the models of global warming and the attention that the “cloud cover” portion of these models has received over the past few years. The dominant models of global warming originally made a fairly simple assumptions about cloud cover — an assumption that may or may not have been valid, and one that the model’s results may not have been particularly sensitive to. However, industry groups that opposed any regulation of greenhouse gas emissions seized on this portion of the model as its Achilles heel.
Meanwhile, glaciers continue to recede in mountains worldwide, global temperatures continue to rise unabated, and worldwide drought has become commonplace. Throughout the entire planet, few people will tell you that their weather patterns have not been changing for the worse in the last several years. Because people are starting to feel the effects for themselves, they now believe that global warming is real. The computer models were predicting such effects at least a decade ago — but because their prescriptions were deemed harmful to the industrial economy, the models were widely ignored or attacked as invalid. (Of course, our current presidential administration continues to ignore these realities.)
This, then, provides one of the central philosophical dilemmas facing a computer modeler. When do you add complexity to a model and when do you rely on simplifying assumptions? Simpler models are easier to control, often provide more intuitive results, and, in many cases, are easier to defend. Complex models represent more of the dynamics of the system and in some cases will provide a better set of results. However, more time and money are required to create and apply a complex model.
The general rule is that you only add complexity to a computer model when you cannot get the model to work without it. Model “calibration” is the procedure of testing a computer model against existing empirical data. If calibration is possible with a simple model, then adding more complexity is not generally desirable. However, what represents an appropriate calibration is again subject to debate. More on that tomorrow.
Today my tasks primarily involve gathering empirical data for such a calibration effort. An arctic high pressure system has made the day crisp and clear, and, like the weather, my spirits seem to be escaping from the doldrums. It’s time to get down to work.
Friday, 26 Jul 2002
Another sunny day in the home office. While I’m not trying to be like a daily Garfield strip, I’m thrilled that it is Friday. (Also, did I mention how fond I am of lasagna and scratching posts?)
This week has been a bit of a grind. Designing and creating an environmental model often involves creativity, vision, breadth of thought. On the other hand, applying an environmental model tends to be slow, thorough, and grueling.
The process of applying an environmental model has three distinct steps: model calibration, model validation, and applying the model for predictive purposes.
In the first step, model calibration, the model is tested against a set of site-specific data to ensure that the model works for the site in question. Calibration also involves varying a model’s parameters to get the model to work as well as possible for that particular site. Many of a model’s parameters have a range of uncertainty. What’s more, many of these parameters can vary from site to site. (For example, the quantity of a particular food source being consumed by a particular predator is not likely to be the same in all water systems.) During the calibration phase, these uncertain parameters are modified within reasonable bounds to see how well the model can be fit to the empirical data.
In the second step, model validation, the model is tested against a different set of empirical data for the same site. This time, though, none of the model’s parameters are changed from those values chosen during calibration. This step ensures that the calibration of the model was not just a lucky coincidence, but that the model actually works for the system given those parameter choices. If validation is unsuccessful, the modelers most go back to the calibration stage again.
Once a successful validation has been completed, the model is ready to be applied for predictive purposes. This allows the user to predict what will happen in the future at a given site or to evaluate different scenarios of site management. This is usually the goal of any modeling study.
This process of model application brings us back to the data and data mining. Good data are required for setting model parameters as well as for forming the empirical data sets against which a model is calibrated and validated. And data, unfortunately, is rarely a clean issue.
For one thing, there are the inevitable units problems. Units are obviously an integral part of the data-to-model interface. Less obviously, they are often the bane of a modeler’s existence. If you didn’t like that part of chemistry where you were required to balance the units in each equation, modeling may not be the career for you. Remember when NASA sent the Mars explorer crashing onto the surface of the planet? Units. That event sent fear into the heart of every modeler I know. (All two of them.)
But anyway, suppose you have a data set that shows units in parts per million. Seems straightforward; your model accepts data in units of parts per million, too. You prepare to bring the data into the model. But wait! There are many, many more considerations to take into account before you start that process:
- Parts of what per million parts of what?
- Was that in wet weight or dry weight?
- Umm. Can we convert wet weight to dry weight?
- What was the sensitivity of the instrument being used?
- Were duplicate samples taken to ensure the integrity of the data?
- Was the location of the data sample identical to the location being modeled?
- Umm. Can we assume that it was close enough?
These types of questions come up time and time again. Fortunately, in many cases, these issues can be resolved after careful consideration. In cases where the data was gathered for the purposes of modeling, most of these issues were preempted by the careful design of the data-gathering procedures.
Notwithstanding the time spent slogging through databases, environmental modeling is still not a bad way to make a living. There are undoubtedly times when I wonder if I should be sitting here alone in front of my computers, tucked away in rural Vermont, given the depth and breadth of environmental and social problems. That’s something to be worked out over the course of my career and indeed the course of my lifetime.
Ultimately, though, I think that it is helpful for us to understand the dynamics behind complex environmental systems so we can understand the effects of our actions. Going back to the example of global warming: If we had no idea why the earth was getting warmer, we’d have no idea what action to take. Now that we know how we’re changing the global climate, we have to push for the political will to do something about it.
Now if you’ll kindly excuse me, I’ve got to go wrangle with some more data issues. At least if I get frustrated I can take it out on my scratching post.