Turning airliners into flying weather stations

OMG the contrail nutters will go ape over this. Just one more piece of “weather war technology” or some such bulldust. I would go a step further and use some surplus UAV’s (Post Afghanistan of course) . Look at it this way. No pilot on-board, can stay in the air for hours without any supervision, can fly above normal air traffic heights and can be tasked away from their assigned area if needed. Just need the right packages on-board, already has the military standard GPS packages installed, could be used to Cyclone/Typhoon/Hurricane/Tornado chase or could be used in a weather emergency to bring back real time weather information.

Oh no ! A Gizmo with occult sensing ! Proof of global wierding !

I haven’t looked at the actual report – do they address the issue that the airspace covered by commercial airliners typically follows well-traveled routes across the Atlantic, which would give an abundance of samples in areas that may have little bearing on predicting the formation and course of tropical storms?
Or perhaps obtaining a large quantity of observations is not the goal, inasmuch as any observations whatsoever over the area of interest would provide data where otherwise there is none.
BTW, as noted above, a lot of the commenters seem not have grasped the concept of the GPS satellite signals themselves being used to probe atmospheric conditions, as opposed to merely being used to give the location of readings taken by more conventional weather measurements.

Regarding GPS locators: General aviation already has the gear and infrastructure, using the Iridium satellite-phone network, generally costing under $200/month to track aircraft position at one-minute intervals. For example http://us.spidertracks.com/
I’m sure commercial airliners could install something similar.

Why not do an ARGO project for the atmosphere? In fact, why not extend ARGO above sea level so that each buoy performs a sounding while on the surface? Why not install a network of fully automated sounding sites all over the planet at randomly determined fixed locations, and actually measure atmospheric conditions on a gradually improving grid that isn’t shaped by the accident of where humans live in large numbers and corrupt local measurements with UHI? Why not install automated sounding devices all over the world’s deserts, where one in principle might be able to directly measure the CO_2-connected GHE change over a decade (at least, if it is as large as it is purported to be, it should be easy to measure there)?
After all, if this were done with random site selection, it would by construction not require a UHI correction at all (and indeed, would give us the data needed to increasingly precisely determine the UHI effect and the mean contribution of LOCAL warming to current estimates of GLOBAL warming).
ARGO itself is flawed because far too many of the buoys are not moored and are not located at randomly selected sites. It’s like nobody has ever heard of Monte Carlo and the importance of random sampling when constructing unbiased estimates, just as people have apparently never heard of rescalable icosahedral tessellations of the sphere and the problems with kridging and extrema on a sparse grid, or using a kridge or any other interpolating polynomial to do a quadrature as a step towards evaluating an estimate of the mean, especially on a coarse grained spherical polar coordinate decomposition of a spherical surface with its polar singularities.
rgb

My only concern is that trans-oceanic tracks are typically moved away from big weather events if at all possible, so that you may not get the information that is of most interest when you could really use it. Airliners are not hurricane hunters; they are hurricane avoiders; the same goes for avoiding storms in the ITCZ when flying N-S or S-N.

Why do gremlins come to mind? ;-(

Katherine says: And after they manage to get the device down to shoebox size, they’ll have to sell the idea to the commercial airlines. What benefit would airlines get from installing it? If there’s none, the device would just be dead weight.

A system regularly transmitting GPS locations would also provide yet another way of tracking of Malaysian Airlines planes. It would be particularly useful if mounted in a place where the pilots can’t turn it off

Ya think?
Well, OK, maybe that was a bit over the top, but they should be aware of all of those things:
http://en.wikipedia.org/wiki/Geodesic_grid
Note scaling in figures — note also that this tiling has been in use since at least 1968 for geophysical problems. Yet it is my belief that basically all of the GCMs use lat-long spherical polar tiling with completely asymmetric tile shapes and areas. I quote from the article (referring to the universally used lat-long grid):
“However, such a pattern does not conform to many of the main criteria for a statistically valid discrete global grid,[2] primarily that the cells’ area and shape are not generally similar; this is especially noticeable as the cells are developed towards the poles.”
As you say, “Ya think?”
Then there is Kriging:
http://en.wikipedia.org/wiki/Kriging
Note well two things from this article. First of all, note the enormous uncertainty of the interpolation in one dimension with a moderately sparse grid. This uncertainty strictly depends on assumptions made about the underlying process being fit — again I quote: “Under suitable assumptions on the priors, kriging gives the best linear unbiased prediction of the intermediate values.” That’s sort of like saying “if the process being interpolated is nonlinear, or if the prior assumptions one makes are unsuitable, kriging will not give you the best unbiased prediction of the intermediate values”. Second, note from the figure itself note that the specific place where the interpolation is quite different from other e.g. spline interpolations is in interpolated segments where an extremum (probably) occurs. Here’s where those prior assumptions come into play — cubic splines are well-known to be very good for certain classes of problem, and still other interpolating polynomials can be good for still other classes of problem.
It seems reasonably likely that kriging underestimates extrema, possibly systematically, and it is virtually certain that it leads to very large errors across the interpolated regions. One is then left with nothing but a “fond hope” that those errors are symmetric and on average cancel, as opposed to systematic so that they don’t.
Finally, there is a well-known connection between interpolating functions and quadrature (numerical integration):
http://en.wikipedia.org/wiki/Numerical_integration#Quadrature_rules_based_on_interpolating_functions
The idea is that one numerically interpolates a set of data points with some smooth function (e.g. piecewise polynomial) according to some criterion, and then integrate the interpolating function one constructs to estimate the integral over some sampled domain. Note well that error estimation is serious business and is based on all sorts of assumptions that any numerical/computational scientist knows can easily be defeated by nothing more than bad luck for any given non-adaptive grid. This is why using adaptive methods are absolutely key.
Note also the comments in the article about multidimensional integrals and the curse of dimensionality. Note the solutions — Monte Carlo (which I mention above) or some very sophisticated methods one can implement for sparse, non-uniform grids on hypersurfaces or in hypervolumes that may or may not be flat. Again, errors are enormously difficult to control for sparse, non-uniform, non-random grids on curved surfaces, and most oversimplified forms of linear interpolation quadrature are going to really suck.
Since estimating global surface temperature is precisely taking an irregular, sparse, non-uniform sampling of points locally averaged in 5×5 degree lat-long surface cells, kriging to interpolate, integrating the interpolation over the area, and then dividing out the total “spherical” area, all of these remarks are entirely apropos, and IMO are serious criticisms of the GCMs right out of the box(es).
I could, of course, be wrong — maybe there are GCMs built on rescalable icosahedral grids — but none of the ones I’ve looked at so far were, and a lot of them (if not all of them) seem to share “DNA” at this level — use common lat-long input data, use a similar gridding, etc.
rgb

spdrdr says:March 18, 2014 at 8:56 pm “How much would it cost … to install a proper GPS locator for all commercial aircraft, which could NOT be disabled by a person or persons unknown? ”
Apparently not much considering that trucking companies and DPW’s already install them in semi’s and snow plows. (However… I believe those rely on ground cell service to rely the GPS info to the home office so airplanes would have to use something like Airfone over the USA unless we start allowing cell phones from airplanes.)
But your point brings up an interesting dovetail to GPS weather reporting from airplanes in that it could easily be incorporated into long-haul semi tractor-trailer trucks right now. There are a lot more trucks, distributed over a lot more area than there are airliners resulting in finer resolution of barometric pressure and temperature, (well.. temperature on asphalt but consistently so for 99%).
Plus, with so many trucks going in and out of cities every day, imagine how it could refine the issue of UHI magnitude?

great idea. SWA is doing soundings out of Love Field here in Dallas now for the local NWS office.