Slashdot Mirror
Floating Computers Keep an Eye on the Oceans
mightysquirrel81 writes "This fascinating picture story shows the tech behind the global Argo progamme set up to monitor the world's oceans. Using 3,000 floating computers and a network of satellites, researchers measure sea temperature and ocean currents to predict climate change."
Yes, they do wash up on shore (sometimes they're even found), and yes, they are continually replaced.
The link is for registered users only. Here is a google search that will lead you to the pdf file. Google result
No, and it wouldn't have worked on Mars either.
The wind energy is already being conserved and eventually becomes heart.
"I've got more toys than Teruhisa Kitahara."
the jury is still out on man made climate change, inspite of what some would force down our throats.
If you mod me down, I will become more powerful than you can imagine....
They drift at depths of 1-2km most of the time, and down there the ocean mostly moves along topography lines so the floats actually tend to float parallel to coasts. Some of them wash ashore, but much less than would occur if they were at the surface all the time.
I'm not totally sure how they do the calibration, I don't actually work with Argo, but I used to work in the same department as one of the main groups manufacturing floats. This is what I remember from papers I've read and stuff I picked up at seminars, so I may have the details wrong. Before they are launched, the sensors are calibrated using a sample of IAPSO Standard Seawater. Once they are out at sea, sensor drift is estimated a few ways.
.1 PSU over their lifetime (Ocean Salinity is generally between 32-36 PSU, depending on where you are) and I think their measured accuracies from the floats they retrieved and recalibrated are within about .2 PSU. So essentially, it's really hard to calibrate these things and they try every method they can think of to do so. 0.2 PSU error may not seem that great, but before the array went in the best guesses we had were probably more like within .5 PSU or so, and we didn't have any information about how it varied with time. There was not a lot of Salinity data from the deep ocean before Argo.
First, if another Argo float moves through the same body of water as a previous one, the measurements they make are compared. Each year you get a few hundred matchup like this and that lets you get an indication of the drift.
Second, there are ship measurements that are taken periodically as research vessels cross the ocean. These measurements can be calibrated very exactly, and they sometimes cross float tracks, so you can compare the ship data with occasional float data. They can also send ships out specifically to compare with certain floats, cause the floats transmit their position home every few days.
Third, they've fished some of the floats out of the water and recalibrated them. The ones they've recalibrated this way are used as a sample to calculate drift rates.
Fourth, the reconstructions of ocean state that they do are based on interpolation and long averages, and if the drift is random, the average of a bunch of the floats should be close to reality. If a float is obviously drifting too far from reality (as measured by other floats or the climatology for the area) the data is flagged and removed from the main dataset.
They tried to make the floats accurate to within about
Similar methods are used to calibrate the temperature and oxygen sensors, but you get the idea.
Very true (I'd mod you up if I had points). More precisely an Argo float generally drifts at 2000m (2000db pressure actually) and then ascends every 10 days to take salinity and temperature profiles and send the data to a satellite. An Argo float lasts about 4 or 5 years after which I suppose the battery is drained and the float may sink or wash ashore.
It's not a cheap undertaking, but the data is absolutely invaluable to oceanographers since it's damn near impossible to sample such a huge body using only ships!
Usefull link: http://www.argo.ucsd.edu/FrHow_Argo_floats.html
I've worked with Argo data on several occasions and I've developed some Matlab code which makes the whole process quite a bit easier. It handles caching/retrieving/querying data from the official database and also a whole bevy of visualization options. It's scripter-only stuff and hasn't been touched in a year or two but was working very well for my needs recently. Query options include date/map polygon/float number/other metadata, and visualization covers a whole range of oceanographic plots from isosurfaces and sections, to property/property plots, waterfall plots, even some protoypical 3d-surface visualization plots... you name it, I've probably done it twice.
Of course, it's all freely available to anyone who might be interested. I only ask that if you make improvements, you share them back so that they can become part of the main distribution.
The Argo dataset is really, really cool and easy to get into! Too bad the resolution is so low and the salinity sensors tend to get fouled over time.
"The wise man proportions his belief to the evidence." -- David Hume