[grecy]

I find it difficult to believe that the large governments, in control of many imaging satellites, could not find something like this if they were so inclined.

Take an enormous number of photos of the region at the right resolution, and have software eliminate the boring (empty) photos and flag the interesting ones (contain something of the right size) for human review. Heck, I bet it could even exclude known shipping lanes to minimize false positives.

[akira2501]

There are 510 trillion square miles of surface area on the planet. Just.. imagine the logistical challenge of taking and storing these pictures, let alone the _process_ of obtaining them. It's almost impossible to achieve anything like a "snapshot" of the surface; because of this, it's almost useless for locating moving objects on the surface.

You could, conceivably, completely miss the object in your imaging passes; or the satellite you have tasked for a particular ground track has an error in it's imaging sensor or downlink equipment, meaning you're missing a huge chunk of data for a while. You might also have to move the satellites around occasionally making imaging useless during these periods as well.

Also.. there's night and weather. You can't always freely image the surface just because you have a camera pointed in that direction.

These are just a small number of reasons that whole-earth real-time imaging is seriously difficult.

[leoedin]

If you're looking for a ship you're going to have much more success with a synthetic aperture radar satellite. They work through clouds and at night, and the bright reflection of a ship's flat steel plates is fairly easy to spot against the background clutter of the sea.

Unfortunately they generally cost hundreds of millions of dollars and there's only a handful of them in operation.

If a western government actually wanted to look for this ship (while it was still floating, anyway), they'd almost certainly find it. There's whole divisions of the air force and navy trained to do exactly that. The cost of the search would be pretty much unjustifiable though.

[jlarocco]

What? You're off by several orders of magnitude. The surface area of the earth is only 196 million square miles. Still very big, but nowhere near 510 trillion square miles.

[david-given]

The Earth's diameter is ~13000 km. Surface area is therefore 4 * pi * r^2 = 4 * pi * ((13000/2)^2) = 500 million square kilometres (which is, as you say, about 200 million square miles).

I suspect the parent is both confusing a million with a trillion and also miles with kilometres.

[akira2501]

Neither. It's just such a habit to type miles. I meant 510 trillion square _meters_, which google and wolfram alpha tell me is correct.

In any case.. the reason I chose that is because that's probably close to a reasonable maximum single pixel resolution for tracking imagery. I wanted the scope of the problem to be obvious by choice of units; unfortunately, I screwed that part up. :(

[david-given]

Been there, done that, bought the aardvark. I mean T-shirt.

(And, of course, it didn't occur to me that 1e6 million square metres is a square kilometre...)

[grecy]

> There are 510 trillion square miles of surface area on the planet.

I understand it's a big job, but that's why big governments have billion dollar budgets and toys. Also remember we have no need to take photos of the entire surface, only the Atlantic (and even then, only the North Atlantic right now)

>You could, conceivably, completely miss the object in your imaging passes

I feel quite confident when images are taken their exact co-ordinates are known, so it can be quickly determined if you've "missed" a patch or if you have them all.

>* or the satellite you have tasked for a particular ground track has an error in it's imaging sensor or downlink equipment*

I feel quite certain they've tested these things extensively, and they're not going find some previously unknown bug while looking for a ship in the ocean. They work properly.

> * You might also have to move the satellites around occasionally making imaging useless during these periods as well.*

I thought most satellites were in orbits that make it possible to image almost the entire surface without moving them.

> Also.. there's night and weather

I didn't say it was going to be easy.

> whole-earth real-time imaging is seriously difficult

I'm not for one second proposing whole-earth real-time imaging.

[akira2501]

> I understand it's a big job, but that's why big governments have billion dollar budgets and toys. Also remember we have no need to take photos of the entire surface, only the Atlantic (and even then, only the North Atlantic right now)

Yes.. but spending a billion dollars to save a few million is kind of silly, isn't it?

> feel quite confident when images are taken their exact co-ordinates are known, so it can be quickly determined if you've "missed" a patch or if you have them all.

Knowing you're missing data is one thing, but if you're _looking_ for a _moving_ object it can really set you back. You also then have to wait for that satellite, or another one on the same ground track to make the next pass. Most earth imaging satellites have a 90 minute orbital period. Even then, the earth is spinning, so you can't get back to the _exact_ spot very easily (if at all). Still confident this is easy?

> I feel quite certain they've tested these things extensively, and they're not going find some previously unknown bug while looking for a ship in the ocean. They work properly.

You can't image from geosynchronous orbit. I mean, you can, but it's pointless. You need a polar orbit. These orbits aren't stable. You need to adjust them (seriously, sign up for the GPS update list and see how often they have to be adjusted and moved).

Also, "they work properly." Clearly you're not an engineer. Things work properly until they don't. Which could be because of component failure, debris strike or you know, plain ole solar flares. These objects are in _space_. It sort of ups the bar.

> thought most satellites were in orbits that make it possible to image almost the entire surface without moving them.

As I said before, these orbits aren't perfectly stable. You need to boost and adjust and make movements. You may also want to put a new satellite in a slot where a previous satellite (with lessor technology) was. This all requires an insane amount of coordination and effort. Or, an insanely huge cluster.

> I didn't say it was going to be easy.

You made it sound as if it was a foregone conclusion. My effort was to point out that it is in no way that simple. It's going to be hard, it's going to cost a lot, you're going to have to throw people and money and equipment at it _constantly_. Is this justifiable for looking for missing objects? Probably not.

> I'm not for one second proposing whole-earth real-time imaging.

Then good luck finding those _moving_ objects. Also, I'd like to know what orbit takes a satellite only over the Atlantic Ocean. Yes, there are orbits that can "park" over an area, but those orbits are useless for the type of imaging you're suggesting, so you have to deal with the whole-earth problem anyways.

[flurdy]

Or use crowdsourced anomaly classification like http://www.tomnod.com/ Though you get more interest in relief, search and rescue cases, not abandoned ships.