January 2009 Archives

Computational Ecology

| No Comments | No TrackBacks
What is a modeler?  What do modelers do?  This is an elusive mystery, wrapped in an enigma, shrouded in ambiguity.  People have been calling me a modeler for years, and I don't even know the answer.

The term "model" is so broadly applied that when a scientist says that he or she has constructed a model, it is almost impossible, without context, to know what that means.  It could be a calculation of sound speed in a dynamic ocean, or of ecological drift in the rain forest.  It could represent the movement of plate tectonics, or of people on a subway.  It could be physical, biological, geological, statistical, etc.  All you really know is that it might involve some kind of calculation or equation--but not necessarily.

Sometimes models get a bad rap.  For example, when the markets crashed last fall, and the global depression loomed, Alan Greenspan explained it by saying,
"I found a ... flaw in the model that I perceived is the critical functioning structure that defines how the world works, so to speak."
When a lot of people rely on a model that turns out to have fatal flaws, modelers all over get a bad rap too, even if their models are nothing like the fatally flawed models.

We can't always help being pigeonholed.  Besides, sometimes it's nice being a modeler, since the term is broad enough to include almost anything.  Plus, we can make puns about "modeling" and "working with models all day long."  For what it's worth, here is a better explanation of what I do.

Most of my work in the EMLab falls under the growing field of computational ecology.  In a nutshell, this means using high powered computational tools and computer science techniques to answer ecological questions.  It involves synthesizing large data sets, working out theoretical ecological and bio-physical relationships, and setting up very large calculations that may take days to compute on our computing cluster.

While this might sound at first a little bit like counting eyelashes, it turns out to (usually) be fun and exciting.  We can apply this powerful field to tasks ranging from protecting endangered whales to understanding the ecological impacts of climate change.  It's a young field.  Universities are only just beginning to form computational ecology groups (e.g. Yale, Michigan State, UC Santa Barbara, et al.), and journals covering the topic are springing up (e.g. Ecological Informatics).  It's also a field with a lot of room for growth, since computing power is always increasing.

Here at the Seascape Modeling lab, the focus is the Gulf of Maine ecosystem.  There are a lot of open questions in this region--some of them fairly controversial.  Along with nets, hooks, buoys, and microscopes, one of the most fruitful tools is a massive series of high-powered computers.

...and, of course, a model:

Mass Bay Forecasts--Coming Soon!

| No Comments | No TrackBacks
 The pervasive cold and darkness that characterizes winters in New England means that primary productivity in the Gulf of Maine shuts down during this time.  However, shallow waters along the coast limit how far below the surface the phytoplankton can be mixed.  For this reason, the spring bloom in the Gulf of Maine starts along the coast and in the south, so biology in Mass Bay tends to lead the rest of the Gulf.  If you're an animal, like a right whale, that can swim large distances, Mass Bay is likely your first stop in your annual tour of the Gulf. Autobuoys.jpg

From the point of view of a right whale, the downside to Mass Bay is that it is surrounded by Massachusetts.  This means that the Bay is one of the more industrial stretches of water in the world.  Large ships bring cargo to and from Boston.  Other ships are supporting the construction of liquified natural gas terminals in the Bay.  Smaller fishing boats move through the area, taking advantage of the same productivity that draws the right whales to the Bay.  Oh, and all those people in Massachusetts, they produce a lot of sewage, and a lot of it ends up in the Bay (after suitable processing, we hope).  The upside of all this activity, is that there is a lot of science going on.  Our colleagues at the Provincetown Center for Coastal Studies regularly collect zooplankton samples, partly to provide information relevant to right whales, and partly to monitor the impact of the sewage outfall.  Our colleagues at Cornell's Lab of Ornithology have installed a series of "autobuoys" that can detect the presence of right whales in the shipping lanes and near the LNG sites (see the image).  All this data creates the perfect environment to try some right whale forecasting.  If we can't do it in Mass Bay, we probably can't do it!

Here's what we plan to do
  1. Run SEASCAPE to estimate the abundance of the whales' favorite copepods: Calanus finmarchicus, Pseudocalanus, and Centropages typicus.
  2. Use the ensemble Kalman filter to assimilate PCCS's weekly zooplankton survey into our model.  This will give an improved estimate of the whale's prey-field and should improve the accuracy of the model over the following week.
  3. Use the model output, plus satellite data and other variables to estimate right whale habitat in the Bay.
  4. Use an assimilation procedure to refine our whale habitat estimates using the Cornell acoustic detections.
I have implemented the Kalman filter procedure and am currently testing it with previous years.  Hopefully, I'll have this working next week and we can start making some forecasts.

SURF'S UP! if you're a plankton...

| No Comments | No TrackBacks

An introduction to internal waves for scientists and future-scientist alike

What are internal waves?
     Internal waves are similar to surface waves (the kind you see and surf at the beach). The biggest difference is that surface waves exist at the boundary between water and air while internal waves exist at a boundary between two layers of water that differ in density. Think of oil and vinegar in the same container: if you let the container sit for a while the oil settles on the bottom in a state of equilibrium (as oil is the denser of the two liquids). You can clearly see the boundary layer between them. Disturbances in this layer (e.g. a drop of oil pushed above the boundary) result in restoring forces trying to bring things back to equilibrium. The dissipation of these forces result in waves that propagate along the boundary between the two liquids. These disturbance-generated waves are just like the waves you see moving across a pond after you throw a rock in and disturb the pond surface layer. In the ocean heavy, cold, and dense water plays the role of the oil in our above thought experiment, and lighter, warmer, less dense surface waters play the role of the vinegar. The boundary between the two water masses in the ocean is called the pycnocline.
     Most frequently internal waves are tidally generated, but they can also be created by underwater earthquakes, significant freshwater inputs flowing into the ocean, the mixing of the water-column by storms, or flow of water over certain bathymetric features.

Why are internal waves important?
    Internal waves provide energy to ocean ecosystems. They induce mixing, and transport nutrients from depth to the surface where they can be utilized by photosynthetic plankton. Internal waves have also been shown to advect (micro)organisms through the water.
What are we studying them for?
     The biota of the ocean are not evenly distributed. In fact there exists a patchiness that is documented throughout the literature which scientists have been trying to understand for quite some time. Areas of high biological activity are often called "hotspots". At a hotspot you might find 40 Eubalaelena glacialis (right whale) feeding on a dense patch of Calinus finmarchicus, or hundreds of seabirds following a pair of Megoptera novaeangliae (humpback whale) diving with a pod of Lagenorhynchus acutus (Atlantic white-sided dolphin) for krill or Ammodytes americanus (sandlance).
      One of the goals of the EMLab is to understand this patchiness. To do this we're looking at internal waves and how they advect euphausiids (krill) as a potential driving mechanism in determining the patchy distribution of hotspots.

All Hail the Kraken!

| No Comments | No TrackBacks
Kraken are supposed to be gargantuan sea-monsters with giant tentacles and a fearsome appetite.  The EML has its own Kraken.  Although our Kraken is small, at least by the standards of sea monsters, it is powerful and potentially fearsome.  Kraken is the name for our computing cluster.  We chose the name because the numerous power cords and ethernet cables emanating from the cluster reminded us of the legendary monsters.

At the moment, Kraken consist of 8 computers:

1 Xserve with two,  3 GHz quad-core processors
7 MacMini's each with one 2 GHz dual-core processor

(the EML is happily PC-free, I'll justify this at some point...).  Multiplying the computers by their number of cores gives us the ability to run 22 simultaneous programs.  Currently, we are using Apple's built in Xgrid system to submit jobs to the Kraken.  When Kraken is treated carefully, with the appropriate deference, it is great for weakly parallel procedures such as optimizing with a genetic algorithm or the ensemble Kalman filter.  When Kraken is not given the appropriate deference (that is, when you screw up the xgrid syntax or how xgrid manages jobs), it can be downright fearsome.

A few of Kraken's tentacles.

Gull's Eye View

| No Comments | No TrackBacks
Lots of scientists look forward to analyzing the data they've collected.  This can be one of the most exciting stages of the scientific method because you start to answer some of the questions you've posed.

For me, the data collection itself can be just as fun.

One of the perks of having cameras that capture images every minute is that we capture many of the beautiful phenomena that take place on the ocean.  As long as they take place on time scales longer than one minute, and as long as they happen when it's not dark out, I'll catch a photo.

The first image shows sea smoke, which seems to happen on a bitterly cold day each year around this time.  When the temperature of the water is warmer than that of the air, and the water evaporates faster than the air can absorb it, the excess water condenses into the smoky apparition that we see.  As long as the temperature conditions are just right, these foggy conditions will persist long enough for us to enjoy the spectacle.


The next image shows some broken ice floating downstream and out of the harbor.  Over the course of the winter, we see a lot of different types of ice on the surface of the harbor.  Some forms originate upstream while others form along the edges of the water.  Taking the time to look at these images and explore the harbor helps to build an intuition about the environment.  To me, it's important to do this before jumping in to the analysis of the data.  I find that taking the time to do this helps me build insight no matter what project I'm working on.


Sea Surface Photogrammetry

| No Comments | No TrackBacks
The image below shows how photography of the sea surface can be used to map surface features.  The photo in the upper right hand corner was taken from the control booth on the Casco Bay Bridge in Portland, Maine.  The image on the left is the georectification of the same photo, superimposed on an aerial image of the harbor.  Using this technique, combined with time-lapse photography, we can study the dynamics of the fluid in the harbor.  We can also trace the movement of visible phenomena, such as sea ice or biogenic oil.

We have collected many months of time-lapse photography of the sea surface, mostly in Portland Harbor.  Our system typically takes a photo once per minute.  We will discuss our results in later entries, and include some cool animations.

Ecosystem Forecasting

| No Comments | No TrackBacks
Beginning in January 2009, we will be generating forecasts for the Cape Cod Bay zooplankton ecosystem.  These forecasts will be produced on an approximately weekly basis, following the collection of zooplankton tow data by the Provincetown Center for Coastal Studies.  These forecasts will include the distribution of patches of copepods--tiny crustaceans that are an important food resource for animals like the endangered North Atlantic right whale.

Stay tuned for forecasts and further details on our forecasting method.


| No Comments | No TrackBacks
This entry marks the official launch of the Seascape Modeling project pages.  We hope to keep an ongoing documentation of the various modeling projects that we have under way.  Many of our projects include operational aspects, such as our ecosystem forecasting, and our sea surface monitoring projects.  The operational nature means that, just as the ocean is changing continuously, our research and knowledge are continuously being updated.  Our goal is to document these updates, as well as our other thoughts and discoveries, on these pages.

We hope that you find our work interesting, and that you continue to learn about the marine environment.

About this Archive

This page is an archive of entries from January 2009 listed from newest to oldest.

February 2009 is the next archive.

Find recent content on the main index or look in the archives to find all content.