SEAMONSTER Basics

From Seamonster
Jump to: navigation, search

Lemon Glacier terminus and location in Southeast Alaska


Introduction

Welcome to SEAMONSTER. This page walks you through what that means in the manner of an FAQ: Where we are, what we're doing, and how we're going about it.


Aside from a droll acronym, what exactly is SEAMONSTER?

SEAMONSTER is science research that motivates and makes use of an environmental sensor web. The project study site surrounds the University of Alaska Southeast near Juneau (red circle on the above map). The project members are University faculty, staff, and students plus collaborators at Microsoft and other institutions. The project duration is three years and hopefully much longer.


What science?

We're interested in understanding our local environment--both physical and biological--as a collection of interconnected systems. Studying that is pretty much the definition of ecology. Each ecological unit--spanning glaciers to tundra to forests to the ocean--can be understood in terms of a lot of interacting organisms and physical and chemical processes. For example the weathering of minerals by sliding glacier ice has a deep impact on the capacity of marine algae to build up sugar molecules from water, carbon dioxide and sunlight.


Could you be more specific about the science?

Sure. Let's begin with a simple question. Is there a connection between stream flow and sunlight? The answer is obviously No. The sun does not create water in a creek. Although come to think of it, the sun could evaporate water up-stream, so there might be a small effect: On a hot day there will be a bit less water flowing in the creek. But that's it.


Now that we have reasoned out the answer to this question let's go ahead and verify it with observation. We put a flow meter in the stream and a thermometer next to the stream. We record both data values for a month (say June 2007) and we plot them on the same chart.


500


The magenta line is the temperature; looks like we had five intervals of warm weather in June 2007. The black line is flow rate. Looks like every time the weather got warmer we got significantly more water flowing in the creek a bit later. Which is not what I predicted a moment ago. So now we have to try and figure out why I was wrong; why there is more water flowing when the sun comes out. (Maybe it's a coincidence.)


Is there more?

To the science: Yes, it never ends. We always have way more questions than we can answer. If you look at that above plot you'll notice that the units of water flow are in cfs which is cubic feet per second. The total water flow in the month of June would be the number of seconds in June times the average flow rate, about 300 cfs. That total volume of water is (let's go to metric) about 1/50th of a cubic kilometer, or about 1% of the ice-mass of the glacier that feeds Lemon Creek. How much of this water in the stream is actually from the glacier versus from other sources? Does this increase in July, August, etc?


So SEAMONSTER is about collecting information?

Yes, but it is also about how we collect that information. In particular can we adopt and adapt current and new technology to get us more raw data (which we like) and can we use wireless transmission to get it sooner (which we also like)? In some cases yes, for easy-to-measure quantities like temperature and stream flow. Some information such as chemical analysis of ocean water requires sample collection and lab analysis. Human observations and photo-journals require a lot of information management as well; so there is no technological free lunch. But technology can definitely help, not just us but perhaps other projects as well.


And once you have it, what do you do with it?

Try and make sense of it, write that up, present the results as journal papers, educational course material, conference presentations, and in discussion with other researchers. We try and include results in this wiki and we also make our data available through an online data server so that others can look at it as well.


Again we are simultaneously doing two things in building SEAMONSTER: Developing and applying technology to acquire information, and interpreting the information to improve our understanding of this environment. The technology part we refer to as an environmental sensor web, described more in the following section.

What is an Environmental Sensor Web?

"Environmental" implies out-of-doors, but we'll describe an Environmental Sensor Web by comparison with an indoors scenario:


Consider a casino. We all know from countless movies that there is a room somewhere with a lot of televisions that are connected to video cameras via cables running throughout the building, and there is a security guard in that room watching them all ceaselessly to make sure nobody is about to commit a crime. Occasionally the security guard might command a camera to swivel to the left or focus in on the blackjack table. But most movies about casinos require that a crime is about to happen and that when it does happen the security guard will be distracted or asleep. Fortunately the video cameras are being recorded continuously so that the police and the unhappy security guard can watch the crime after the fact. Unfortunately just as often one of the criminals cuts a critical cable and they lose the video from that crucial camera.


An Environmental Sensor Web is just about the same thing with some substitutions: The security guard becomes a group of scientists, teachers, and students. The video cameras become, well, video cameras, but other electronic sensors as well: Temperature sensors, water chemistry sensors, rainfall sensors, wind sensors, seismic sensors, underwater microphones, GPS receivers floating around on buoys or anchored in glacier ice, and the list goes on. Instead of cables running through the casino we use radios to send data back to the control room, which is a group of computers connected to the internet. In this way the students can see what is going on in this watershed in Alaska even though they may live in Indiana.


What about the criminals? In Southeast Alaska we do have to deal with criminals, who look like this...


BearOnBeachJuly04 2007.jpg


They may look cute but they're wicked thugs. Look what they did to our nice meteorological station...


MetStationBearDamage.jpg


Let's face it some bears have too much time on their hands. Paws. Whatever.


Motivation for working on sensor webs

Across this very small but pleasant planet there are a upwards of a few hundred thousand people who study the environment full time. Often this involves traveling to remote locations and camping out for as long as possible, observing say a volcano or a rainforest or an ice cap. The result is information intended to help refine and improve our understanding of the details, the guts of how the system works. Sometimes when it is time to leave the field the scientist will think "Too bad I'm out of food; I wish I could stay here with these instruments for another few months of data collection..." The motivation for Sensor Webs: Let the scientist go home but she gets to keep her instruments out there taking more data.


For some time now scientists have been using devices called data-loggers to record sensor data autonomously. What we maintain is: The technology has developed that enables doing much more than this, in terms of cost, data recovery time, health monitoring, and autonomous responses to changing conditions in the environment being observed.


Lemon Glacier lake: Early summer Same location: Late summer

How do you build sensor webs?

First collect together ideas about what you want to do in terms of a scientific inquiry. I began with the idea of carefully watching glaciers flow because they have some bad plumbing underneath them that affects how they move and I wanted to understand this better. A colleague of mine wants to listen to sound echoing around in glaciers (they make a lot of internal noise plus he has a big supply of dynamite) and so we got together and invented a device that could do both jobs for a couple of years at a time, regardless of how cold it got outside. The problem is that inventing something and getting it to work are two different matters, so it has taken awhile to get this device (called a microserver) working, and it is still in development and testing.


A microserver is one example of the type of hardware that can be applied to our general problem of better understanding the environment. It is a rather power-intensive and expensive device and so we also turn to lighter-weight devices called motes to do simpler tasks and communicate over shorter distances. For less cost per unit they can be used to give a spatially dense sampling since you can afford to build more of them on a fixed budget. And for $20 we can go one level lighter still: Temperature data loggers that look like watch batteries.


So this in short is a top-down approach: You create some scientific objectives and you look around for appropriate technology and you build an appropriate deployment plan. Then you build the devices and place them out there and you begin collecting data.


You can also take the approach that you want to build a generally useful device for these types of projects. So you get a bunch of components (computers, radios, software, GPS receivers) and figure out how to wire them together. This is an engineering integration task and returns us to that first idea of microservers. Once the components are working together you can worry about how to provide the resulting device with enough power to make it run for awhile, how to protect it from the elements and so on. You write the operating software for the computer or computers involved and then you can begin testing, eventually in the field where your designs and the science objectives mesh together. The point here is that you are putting a computer in the field which provides a considerable amount of functional flexibiltiy.

Where do you build it?

You build the pieces of the network in a warm dry laboratory somewhere, preferrably one with a nice coffee machine. Once you think it is working you re-build the operational sensor web in an interesting environment. Perhaps this is your back yard, or a stream 3 miles away, or perhaps it is on the sea-floor where Krakatoa used to be. SEAMONSTER is built into the mountains and coastal marine environment near Juneau Alaska. Other sites for developing this technology, from an initial emphasis on polar regions, have included:

  • Greenland: Motion of the Jakaobshavn Glacier
  • Devon Icecap in the Canadian Arctic: Motion of the Belcher Glacier
  • Columbia Glacier, Southcentral Alaska: Seismicity from glacier calving (and dynamite)
  • Antarctica... to be continued

In addition to expanding SEAMONSTER in Southeast Alaska we would like to go to a other places of interest: A game preserve in Laos, a watershed in the Gambia, watersheds and mines in the Colorado Front Range, volcanoes in the Pacific Rim, some acid mine drainages in the Western United States and Romania, and so on. Here is some elaboration on sensor webs that would be interesting to build.

How much does it cost?

You can build a pretty decent small sensor web for about $2000 plus some labor. The resulting system will produce about 30 data streams, will operate for four months continuously, and can be anchored by radio to a laptop computer (cost not included). This page provides details.


That's an off the cuff answer for a hydrology study. A correct answer will of course depend on what you want to do. In more general terms: There are roughly five costs in building and using a sensor web:

  • Hardware
  • Development and building effort
  • Deployment
  • Maintenance/Clean-up
  • Analysis


Hardware costs are not huge until you want to build many units. (Microservers could run between $300 to $4000, motes perhaps $50-$200.) Development, on the other hand, can run to money. As noted above, a big motivator for me in taking on this work has been to solve integration and development problems to reduce the cost of development. It would be ideal to get things so "bullet-proof" that there would only be hardware and deployment/clean-up costs incurred on an investigator's budget (not including analysis of resulting data).


What is this "SEAMONSTER Wiki"?

A wiki is just a web site, a series of interconnected pages, which is really fast and easy to edit. We use a wiki to document SEAMONSTER so that many people can add content simultaneously. There tend to be two types of web pages here (although they all use the same basic format). The first type of page is a working page that has notes that are jotted down in the process of figuring things out. If you land on a page and it doesn't particularly make much sense: That's a working page.


The second type of page is a polished page that has been carefully edited to provide somebody else, perhaps you, with what we hope is useful information. This page is an example of a polished page as is the Main Page (but of course any piece of writing can always be polished further). Polished pages tend to have pictures or diagrams embedded in them and they may indicate that they are polished by using slightly larger font.


Some ideas about wiki formatting can be found here.


Where should I go next?

To explore this wiki you can't go wrong by starting back at the main page site map.


Where are the robots??

SEAMONSTER is a sensor web and a robot can carry sensors around, so are there robots in the SEAMONSTER project? The answer is: Yes, or nearly so; we're getting there. We do have wiki pages about robots and robotics. Robotics are big with tech hobbyists now (for encouraging developments take a look at the Microsoft Robotics Studio site for example). And we've begun collaborating with Georgia Tech on robots built into snowmachines. So the robots are coming and that's pretty exciting.