Recently in Ecosystem Forecasting Category

The concept for this entry comes from a presentation by Jeffrey Runge.  While practicing a talk at GMRI in front of a small audience, Jeff brought up an interesting analogy for modeling.  He explained that modeling is like impressionistic painting.

Some time ago, I wrote an entry philosophizing on the idea of an eco-cast -- that is, an ecosystem forecast delivered like a daily weather forecast, complete with a debonniare newscaster.  This ... was my vision:

Sadly, our copepod / right whale forecasting project has all but wrapped up by now.  We did produce a few forecasts over the years (e.g. here, here, here, here...), both online and in print, but we never did the movie-star version.

A couple of days ago, on a ferry ride to work, I threw together a whimsical video of what such an eco-cast might look like.  The quality (including the newscaster) is not what you would find on cable TV -- you'll notice the amateur nature of it right away -- but it does start to make tangible the idea of an eco-cast.

Enjoy.

-Nick Record, signing off

We can measure the speed of light to an astonishing level of accuracy.  Uncertainty is a millionth of one percent.  We can measure and count sub-atomic particles that weigh a billionth of a billionth of a billionth of a gram.  We can detect earth-like planets hundreds of trillions of miles away by measuring wobbles in stars.

Yet we can't measure egg mortality in the ocean.

This week I'm at the Ecosystem Studies of Sub-Arctic Seas conference.  I've attended this meeting, and other similar meetings, in the past.  It's a good chance to meet with scientists from all over the world, hear their ideas, learn from them, and (if you're like me) interact with them awkwardly.  It's fun and educational.

Learning about the North is especially interesting, as the Arctic appears to be a harbinger of what's to come with climate change.  Complex systems of ocean currents and ice are shifting into new states, and consequently, so are the ecosystems they support.  Marine scientists are intrepidly probing the depths to help us figure out just what is happening, and how to deal with it.

The problem is that the sea is an unknowable beast.

One of my favorite talks at this meeting was on copepod egg mortality.  The question boils down to: how many eggs survive from one day to the next?  After showing measurements made using a variety of different techniques, the take home message was: none of these techniques works.  Actually, I saw a few talks on the impracticability of measuring mortality in the ocean.  Nobody seems to be able to do it except in highly simplified or idealized cases.

It's amazing all that we've accomplished in science.  I marvel in the feats of humanity against the powers of nature, staving off death and disease, conquering flight and space travel, and discovering the imaginary numbers.  That does give me some hope that we might have the cognitive toolkit and the sheer determination to figure out the science of changing seas and their ecosystems--to buffer humanity against the oncoming climate shifts.  But I marvel too in what we have yet to accomplish.  We haven't been able to travel back in time; we haven't been able to build cities on Mars or crack the speed of light.  

And we can't count how many eggs die in the ocean.  One might say oceanography is harder than quantum physics.  At the least, it's an intriguingly mysterious enigma.

-Nick Record, signing off

A fun part of being a grad student, is making bonds beyond the regular "have-to-share-the-same-open-area-office" friendship. The challenges to overcome are so tough, the emotions shared are so strong than you can never break those bonds. So last week was emotional for me, as I assisted to the Ph.D. thesis defence of my last two buddies from my grad school modeling lab.

Both worked on the Hudson Bay system, a very exciting environment to work on. It's the southernmost Arctic sea, a transition zone between the Arctic Ocean and the Atlantic Ocean at the forefront of the impacts of the current global warming.

Pierre St-Laurent defended brilliantly the 17th of May his thesis entitled "Variabilité saisonnière et interannuelle des eaux douces dans les mers Arctiques : Le cas de la baie d'Hudson".

Pierre showed the audience how the fresh-water budget is regulated in the Hudson Bay. He tackled both liquid and solid (seasonal sea-ice) aspects of it. As an example of how great a tool is modeling in a well formed scientific mind, he first studied this issue with a realistic high resolution sea-ice / ocean 3-D circulation model of the Hudson Bay, developed in the numerical modeling lab of ISMER in Rimouski.

He then constructed an idealized system to sort out the relative importance of the various hydrological, atmospheric and oceanic forcing.

This allowed him to demonstrate for the first time the role of changing wind regimes in the periodic retention/expulsion of fresh water from the Hudson Bay towards the North-West Atlantic.

Pierre will soon lend his brain as a post-doc to the Old Dominion University in Norfolk, VA

(He's too modest to agree for me to tell you that there is a tenure track position attached at the end of his 3 years as post-doc).

Virginie Sibert defended not less brilliantly the 20th of May her thesis entitled "Modélisation de la variabilité saisonnière et de la sensibilité au climat des productions glacielle et pélagique de la baie d'Hudson".

Virginie managed to build a model of primary and secondary production within the sea-ice in Hudson Bay.

She coupled this to the ice compartment of the same high-resolution 3-D circulation model than Pierre. After characterizing the spatio-temporal patterns of this system, she coupled it further with an NPZD pelagic production model to have a complete picture of the primary production in the system.

View image

After a rigorous validation process which guaranteed a good confidence in the model results, she finally tested one of the IPCC scenario of climate change (A2) for the Hudson Bay system.

A nice outcome of her work is that the Hudson Bay system would not, for its most part, pass a tipping point yet. Primary production of both ice algea and phytoplankton would increase, even if their respective blooms would occur sooner in the season.

Virginie has already brought her talent and charm as a post-doc in the IFREMER lab of Brest, France.

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:

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.

Andy's concerns about discriminating wisely between trend and noise, between low-frequency and high-frequency signals in time series of environmental variables (Air temperature in his example), apply as well to measurable quantities in ecosystems. Particularly relevant is the phenology of the species, which defines the timing of crucial recurrent events of their life cycle, like the date of first arrival to the nesting ground, the date of germination, the date of mating, the date of blooming etc. You may have noticed that I used examples related to birds and plants. Well, while phenology is a general concept, an empirical knowledge patiently accumulated by bird lovers and Sunday gardeners of the 19th and 20th centuries was first to be translated into systematic scientific surveys. And following rigorous statistical analyses of those long time series of observations, it has been firmly established that changes in the phenology of most species accompanied trends in temperature. The strength of the correlation is all the more important as the seasonality (latitude) of the ecosystem and the dependance of the species to their environment increase. The connection with climate change issues is straightforward. And it is not just about how one species or another will cope with changes of its environment, but rather about the interlocked interactions between all those species.

If you can easily think at a beautiful tulip as a species embedded in its environment, it is the same, in a more dynamic way, for planktonic marine species. Now oceanographers begin to benefit of the fruits of several long lasting monitoring programs. Unfortunately, the ever increasing pace of global climate change means that oceanographers are required to draw firm conclusions about the impact of environmental variability on ecosystems and develop predictive capabilities in the meantime ! And this will remain an elusive target as long as the mechanisms gearing those changes are not understood properly. Daunting task, as the changes in timing of such major event as diapause entrance and exit emerge form several layers of physiological and behavioral processes obeying their own dynamics while interacting with each others. But impossible is not known at the EML, so we decided to model the mechanisms behind the diapause of the dominant copepod Calanus finmarchicus. We already know that even if it can produce several generation a year, this critically important species thrives in its seasonal environment (Northern half of the North Atlantic) thanks to its diapause strategy, which means killing time at depth in order not to be killed by the detrimental conditions prevailing at the surface in winter. For this purpose, it makes a feast on large phytoplankton cells (mainly diatoms) during the short period they are available, and build up impressive amount of energy rich lipid reserves. Those swimming droplets of lipid are in turn the basis for the rest of the upper trophic levels.

And what about changes then ? Things are more sparse there... Records of physiological properties related to the diapausing strategy are about a decade old now. Not enough really to study trends on climatological scales, but enough to understand that interannual variability is high (see figure). But abundance data are enough to see changes, especially in areas localized at its biogeographical fringe. In the North Sea for example, the ecosystem shifted from a copepod population dominated by 80% of C. finmarchicus before the 60's to a present state dominated by 80% of its southern congener C. helgolandicus. What is the role of diapause in that ? Not known yet. One thing is certain though: changes occur at an ever accelerating pace, and the unforeseen consequences for the ecosystems are likely to appear before our eyes while we are still racing to improve our understanding. I strongly wish that Copenhagen "talks" will end up with agreements as legally constraining on our leaders than the climate changes will be actually constraining on us.

Superimposed to the climatological (2004-2008) relative abundance of the different copepodid stages are box plots of the estimated dates of initiation (late winter) and termination (summer) of diapause in Calanus finmarchicus in the Gulf of Maine. Data from UNH COOC WB-7 station. Figure from James J. Pierson.

Weather forecasters enjoy (or lament) the gratification of finding out how good their forecasts are just a few days after making them.  Ecosystem forecasters, on the other hand, often have to wait a long time before we know.  One of our forecasts is for the arrival date of right whales in the Great South Channel critical habitat.  Each year we try to predict when the whales will arrive in large numbers to feed there.  Then we wait, sometimes for years, for the data to come in, so that we can see which predictions are correct, and which need to be re-examined, and why.

We now have the right whale sightings data from 2007 and 2008, so the moment of truth is at hand.  I haven't dug into the analysis yet, but here is the first cut.  Our 2007 prediction was very close, falling just a few days from the actual arrival date, and falling well within the predicted window (see figure).  Our 2008 prediction was later than the arrival date.  This appears to be driven by a period in mid March when there were lots of whales, followed by a lull.  Use of the habitat picked up again in mid April, and persisted from there.  Our model appeared to pick up on this later arrival, missing the brief but intense party in mid March.  It'll be interesting to dig into this further when we start our analysis, and see if we can shed light on what was happening in mid March.

These results are Dan's, but he's indisposed and the results are too cool not to blog.  Dan fit a MaxEnt model for right whales in Cape Cod Bay based on hindcasts of Calanus and Pseudocalanus for 2003-2006 as well as SST, chlorophyll, and bathymetry.  He then applied the habitat model to the 2009 copepod estimates and satellite data to predict right whale habitat.  I extracted whale sightings from the PCCS reports and plotted over his images.  By and large, the PCCS sightings fall within areas of high habitat suitability (5/18/09 Note: figures with sightings have been removed).  The correspondence will probably (hopefully?) improve if we produce habitat maps for the specific survey time, rather than the nearest 8 day image.  Now, here's are forecasts for 5/9 and 5/17:

Our copepod forecasts are now appearing in habitat assessment reports produced by the Provincetown Center for Coastal Studies.  The PCCS runs cruises approximately weekly to characterize the prey resource for right whales in Cape Cod Bay.  Our forecasts include their samples from the previous week, coupling them with physical data to project into the upcoming week.

Here are a couple of our forecasts, with comparison to the actual data collected around the same time.

This plot shows a forecast for April 11th, for total copepodid zooplankton in the bay.