Seascape Projects

Here's a look at the poster I put together for the Ocean Sciences conference:

I tried to keep it consise since so many posters look like the researcher barfed text all over them.
Also, the pictures allow a passerby to quickly get a sense of what's going on, and then, if they're interested, they can ask questions.

Feel free to post questions in the comments!

Below is a link to a pdf of the poster, but you should also be able to click on the image above to see a larger version.
stetson_OceanSciences2010_vFinal.pdf

This post is a follow-up to my ecological forecasting post last month.  The work that I presented at the Ocean Sciences meeting was built around the idea of producing forecasts of ecosystems.  Much of what I presented was discussed in that post, so I'd just like to take a little space here to build on the vision.

I think about ecosystem forecasts as close analogues to weather forecasts.  There are, of course, important differences, but I'll save that discussion for a later post.  Here I'll focus on the analogy.

Early weather predictions were based on signs and signals from the immediate surroundings, and on past observations.  These observations were interwoven with ecological ones as well.  People have probably been making these sorts of predictions since the first homo sapiens scratched their heads and gazed upon a bold red sunset.  Unfortunately, these first people left nothing to cite, so we'll have to do what we usually do, and refer to Aristotle on the matter.

Aristotle included many guidelines for weather prediction in Meteorologica, but actually my favorite work on the matter is Theophrastus' The Book of Signs.  For centuries, predictions were made by following signs like this one, taken from The Book of Signs:

"It is a sign of rain when a tame duck gets under the eaves and flaps its wings."

In the 20th century, scientists began using computers to make forecasts.  In the 1950s, the first computational forecasts were inferior to the more subjective, duck-watching methods.  Computer models would suddenly predict an army of cyclones marching across the country.  You would probably produce a better forecast by just saying, "Tomorrow's weather will be pretty much like today's."

Yet there was intense optimism regarding the potential of computational prediction.  John von Neumann, one of the great scientists of the century, stated:  

"All stable processes we shall predict.  All unstable processes we shall control."

This was followed by decades of steady improvement.  Computational weather forecasts are now the norm.  While we do complain when meteorologists get it wrong, I think it's clear that today's forecasts have immense value and are a crucial part of how our society operates.

Von Neumann's vision, however, has not been realized.  One contributing factor was the discovery of what is termed "chaos".  Without going into too much detail, I'll just say that mathematicians discovered fundamental limits on the predictability of certain mathematical systems, including weather and ecosystems.  As the founder, Edward Lorenz, phrased it:

"Predictability: Does the flap of a butterfly's wings in Brazil set off a tornado in Texas?"

Nevertheless, with better and better computers and techniques, scientists continue to improve weather prediction.  A testament to their utility is the ubiquity of weather forecasts in our day to day lives.  This history provides good context for the development of ecosystem forecasting.  With steadily improving models and increasing monitoring, we are poised to transition from the more subjective and sign-based forecasts to more precise computational forecasts.  Naturally, it will be some time before we're forecasting at the level of meteorologists, but one day in the future, you might be watching a broadcaster like this on your daily news:

The coolest part of the media experience has been watching the story spread around the world.  This is a testament to the global reach of the BBC as well as the global interest in carbon issues, especially in Europe.  As of 8:30 EST on 3/1, stories based on the BBC report have appeared in FranceSlovenia

Note:l my ambition to blog each day fizzled like a Portland (OR) rainstorm.  We'll try to add some additional reports from the meeting in the coming days. 

For me, the most memorable part of the meeting was being invited to do a press conference--something I've never done before.  I was invited to give a regular science talk in a session on the impact of climate change on marine ecosystems.  I thought I would use this as an opportunity to talk about some calculations I've done characterizing the carbon footprint of whaling (see this earlier post).  AGU, one of the societies that was running the meeting, thought the news media would be interested in this topic.

The hardest part was deciding to do it.  Since I hadn't presented my calculations to many other scientists, I was worried that there was something I was overlooking.  Visions of cold fusion were dancing in my head.  In the end, I decided to go for it.  To prepare, I organized a mock press conference at GMRI, with my colleagues acting as journalists.  This was extremely helpful.  At the conference, I spoke for about 15 min:

and then took questions.  In addition to the reporters in the room, there were a couple joining on the phone.  I then spoke with several reporters one-on-one, including the BBC:

The BBC story was online by 11PM (PST) last night, and by this morning, it had been translated into Hungarian, Slovenian, and Italian (I didn't know I was fluent in Italian).  Here are some links to a few of the stories, if you're interested in reading more.  All in all, a really fun experience.

BBC

Environmental Research Web

Discovery News

The official meeting started at 8AM this morning.  Meetings like this are the intellectual equivalent of drinking from a firehouse.  At any given time, there are 15 different sessions in progress.  Each session is organized around a particular theorem, and the themes at this meeting cover the full gamut of oceanography.  About a year ago, groups of scientists submitted proposals for sessions.  Once the sessions were selected, the oceanographic community was asked to submit abstracts.  An abstract is a brief (~one paragraph) description of a study, and when you submit an abstract, you select which session you think is most appropriate for your work.  Then, one of three things happens.  1. The session rejects your abstract, possibly passing to another session, 2. The session accepts your abstract and invites you to give a talk, or 3. The session accepts your abstract and asks to you prepare a poster.

Usually, breakfast is spent looking over the titles of the talks, and figuring out which ones you'll try to see.  One talk was easy to add.  Our very own Frederic Maps gave a talk at 8:30 in the morning on his work modeling copepods.  Depending on the talk and the session, you can have anywhere from a few people to more than 50 (remember, you're up against 14 other talks).  As you an see from the picture below, Fred's talk was quite popular:

The meeting officially starts tomorrow (Sunday) night, so why did I fly out on Friday?     Science has always relied on communication and collaboration--hence, the need for conferences.  Oceanography is an inherently interdisciplinary field, but it is very hard to get truly interdisciplinary projects funded. One way to get some interdisciplinary work done is to organize a workshop.  The idea behind a workshop is to get a few very busy people to take a few days from their day-to-day work in order to work together on a common problem.  So, this is why I'm spending this weekend in a conference room instead of hiking with my family.

The point of this weekend's workshop is to develop a better understanding of how changes in the Arctic affect the North Atlantic.  I've stumbled into this line of research by uncovering a dramatic change in the Gulf of Maine plankton community that took place around 1990.  Turns out, lots of other things changed right around that time: the waters became less salty and began flowing faster, herring became more abundant and right whale calves became rarer.  Many of these changes were observed from New Jersey up to Newfoundland.  The best explanation so far is that these changes originated when the winds over the Arctic pushed a slug of fresh water and ice into the North Atlantic.  This created a pocket of fresher water that eventually moved down to the Gulf of Maine:

 The conditions that created this slug persisted through much of the 1990s.  The workshop, organized by my colleague (and former Ph. D. advisor) Chuck Greene, has brought together biologists like me, physical oceanographers, and Arctic climate specialists to try to get a better understanding of exactly what happened. 

The Ecosystem Modeling Lab is headed to the other Portland for the biennial Ocean Sciences Meeting.  OSM is the main oceanographic conference.  Conferences like OSM are an important part of science.  They provide an opportunity to learn about the latest developments in the field, catch up with colleagues, and find collaborators, employers, students, and post docs.  This week, I'll try to give an inside view of  a scientific conference.

First up, getting there.  I'm writing you from a tiny little desk in the middle of O'Hare airport.  I find that I can often get a lot of work done while traveling.  Here's me working on my flight:

On the flight, I managed to update a time series of temperatures from the Gulf of Maine and do some analysis of the relationship between the number of right whale calves and the amount of their food.  I walked off the plane feeling good, but with a battery at 50%.  I've become pretty good at finding power outlets in airports, but O'Hare seems to be doing a good job hiding them.  They do provide some very tiny desks with power outlets.  Between the small desk, uncomfortable stools, and exposure in the middle of the concourse, I think I'll be moving on.  

Early numerical/computational weather forecasts could not compete with traditional forecasting methods.  Simple lore-based indicators--like "red sky at night, sailors delight"--could, in many cases, out-perform computational models.  In fact, simply predicting that tomorrow's weather will be just like today's was much more reliable than the cyclone-ridden numerical predictions put forth by scientists.

Over the past 50 years, computational power, theory, and data collection have improved, leading to a revolution in weather forecasting.  What began as an inferior  practice quickly overtook other methods.  Computational predictions have now replaced traditional subjective predictions, and detailed weather forecasts are integral parts of our everyday lives.  The figure below (adapted from Shuman 1989) shows how computational weather forecasting models have progressed since the 1950s.


Figure adapted from Shuman 1989.  Error in numerical weather forecasts has decreased substantially since the 1950s due to improvements in the models and the computers.

In contrast to weather forecasts, there are very few ecosystem forecasts available.  Those that are available are at the cutting edge of science, and are often released only after the passage of the events they are trying to predict.  Yet I would argue that we could be on the cusp of a similar revolution in ecological forecasting.

Many lessons from the past 50 years of weather prediction can be carried over to the development of systems designed for ecological forecasting.  There are three key components to a good forecasting system, all of which we can begin to implement and/or take advantage of right away.  These are: monitoring (steady streams of data), adaptive computational algorithms, and forecast availability.  The common service provided by all of these components, and the crux, I would argue, to a revolution in ecosystem forecasting, is feedback.

I've sketched out below a concept map of an ecological forecasting system, highlighting the mechanisms of feedback.  Briefly, data is input to the algorithm, which produces a forecast.  This forecast then becomes available to a number of feedback avenues, including the designer, users, and the algorithm itself.  (Side note: I used the "Concept Map Builder" designed by the Center for Ocean Sciences Education Excellence, which is a tool under development, worth checking out.)

Monitoring is, of course, a critical component as it provides input data.  In this system, however, data plays another role.  Suppose we are producing weekly forecasts.  Then each week, we want to know how well we did with our forecast.  At first, our forecasts might be little better than a random guess.  That's okay.  As long as we have the data coming in to tell us how well we're doing.  When we're not doing well, we want to know so that we can adjust.

That brings us to the second component: adaptive computational algorithms.  Using techniques borrowed from computer sciences, such as the genetic algorithm, a system can learn based on its past successes and failures.  It's important for the system to be able to assimilate a steady stream of data and to adapt based on that.  Ideally, a system should be flexible enough to incorporate new theoretical information as well.

The third component, forecast availability, is one that has been missing.  In my opinion, this has been holding back progress in ecosystem forecasting.  I suggest that, like numerical weather forecasters, we should be putting our forecasts out there, even while they are little better than subjective forecasts.  It's okay if many forecasts fail.  With an ensemble of forecasts available, and with a steady stream of observations coming in, our computational systems can begin to evaluate, learn, and adapt.  Over time, the failures will presumably be fewer and fewer, and the forecasts will be more and more useful.

There are a number of subtleties that I've brushed over here, for the sake of simplicity and generality, as well as some practical considerations that are nontrivial.  Nevertheless, these are the basic ingredients, as I see it, to a revolution in ecosystem forecasting.  I'll be presenting these ideas at the Ocean Sciences meeting in Portland, Oregon next week.  If this sort of thing is your bag, look me up.

Just a quick note on our sea surface monitoring project. We are working with a group in Ensenada, Mexico to apply our camera system (designed for oil spill mitigation) to a red tide monitoring project. The images below show a dry run, so there is no red tide present, but stay tuned. If this project gets off the ground, it would be a neat application of our system.

Original photo