Cathryn asked some good questions several threads below....one of them was whether GPS could have provided early warning of the tsunami. The answer is "yes", if the systems were set up correctly, in the right places. This was probably not the case this time, but could be the case in many places the next time....but it is not the cheapest solution....it just may be the best. First, I heard several days ago on a news report that GPS had shown that there was 20 to 30 m of motion of (I assume) NW Sumatra and/or Nicobar islands. I was curious how they knew this so fast (I heard it 2 or 3 days after the quake). I was told that the French might have continuously-recording GPS in place before the earthquake.

In the NPR link in the post above, it was discussed that it took quite some time before they knew that the quake was as large as 8.5 (I think I heard there or elsewhere that it was almost 2 hours). If there were 4 or 5 continuously recording GPS stations distributed along the west side of northern Sumatra, the Nicobar, and Andaman Islands, powered by batteries, recharged by solar panels (or whatever), sampled every minute or more often, and communicated (by satellite?) to a center where they are automatically analysed by computer...with maybe trained analysts on hand 24 hours/365 days...then within maybe 5 to 10 minutes the motion would have been seen....and it would have been fairly certain that, given this motion and the large area over which it occurred, and knowing the position of the faults...especially the main subduction zone, that there almost would have to be a large tsunami...if you shove against 1000 km of ocean by as much as 20 to 30 meters, a tsunami will be produced. A general warning could then be issued. But, that would not be good enough. The information from continuous GPS would have to be fed into software for some scenario earthquakes, along with tsunami-modeling software that makes use of the depths everywhere in the affected ocean. That could be done quickly. Then, there would have to be communications in place to warn all the affected coastal areas. And, the populations would have to have been already educated on what to do once an tsunami warning was issued....

But, if you had only the continous GPS and "real time" analysis of it and not the rest of it....you would still issue a general tsunami warning.

As discussed in the media, given all the other problems that poor countries have, a system of this quality and extent was not something that would be done for an ocean that had no historic record of a tsunami of this scale (Krakatau tsunami was not nearly as large once you were 1000 km away....).
Chris