Creating a Tsunami warning system today can leverage the networks ( and robust routing ) to combine networks of sensors with distributed recognition of the danger zones, use of computer models of the ocean basins, to get multicast warnings to folks, to radio stations, to alarms where exists, to use weather radios, text messages, email, and satellite broadcasting.

Seismic networks knew within minutes that a great quake occured, and the location was plotted so a source term was available. Within another minute if the Tsunami computer models were ready, a plot could have been generated that would indicate regions of concern, and approximate times. Certainly having data buoy reports from the area of the quake would improve accuracy, but that would be a refinement that may be unaffordable or unavailable until later.

This simple model, namely using existing seismic monitoring networks coupled to a server running a basin model would yield preditions that could be rapidly distributed to relevent authorities, news organizations, and the like. Model based predictions could also generate validation tests. The sort of events that cause a tsunami can also be heard through the networks of hydraphones that are listening all the time, tracking subs, etc.

Any tsunami wave that would be generated would also generate acoustic signatures as it interacts with the coastal areas as it sweeps away from the epicenter of the wave generating event. This signature would validate the predictions. Satellite data collection for the fixed bouy system is essential so that the tsunami generating event doesn't just wipe out the system. With a world wide distributed system, it is likely that the notice of an approaching wave would originate from outside the affected area, and could thus be more effective.

Clearly credibility requries that there be humans in the system to ensure the integrity of any warnings. We really should be doing this in California, and certainly in Florida. Here in California we have a Tsunami warning system, though the delivery of the messages would be unreliable at best. I remember the Cresent City Tsunami warnings, on KNX in L.A. although I don't remember the specifics. The Cresent City event caused folks to be wary in Long Beach, as there was no network at that time ( my recollection anyway).

The U.S. Navy has such models of the ocean basins, and the software to generate the maps predicting the path of a tsunami. Without giving away classified information, the U.S.Navy could provide the alarm information in a timely manner. The U.S. Navy is also the best qualified organization in the World to provide the validation as they operate the largest network of monitored hydraphones, have the most sophisticated signal processing software.

The U.S. Navy would probably need some time to write some computer code, and maybe more time to come to grips with giving out formerly secret information, but they could provide value to the people of the world in one authentic basin wide tsunami warning that would pay for the entire cost of the Navy over our history!

Internet based warning could take seconds to reach decision makers in various countries, and once the satellite broadcast media picked up the warning, large populations at risk could move to safer ground, mitigating the loss of life to a major degree.

Time to get this in place for the globe. We should have monitoring and warning protocols in place for every ocean basin. A meteor could also generate a tsunami, and would likewise be observed by the U.S. Space Command, though I doubt that the Navy and Space Command have working interconnected systems for this purpose at this time. Never the less, such a connection would be useful, and it should generate civilian warnings in the likely event that they are needed some time over the next 100 years. It would be very inexpensive to put such a system based on existing sensor networks on line compared with the much discussed special purpose systems using only governement assets.

Volcano's that generate Tsunami's are likewise detectable by seismic stations and could be confirmed by existing Naval surveillance systems, generating the same internet notifications, news bullatins, etc. The threshold for such notifacations should be set quite high, so that the confidence in the warning would justify the extreme measures necessary to cause a mass evacuation, justifying the inevitable casualties from the evacuation in the reduction of overall harm.

Now with some 100,000 prompt dead, perhaps many times that in disease to follow, and another challenge to our civilization. Civil disorder in Sumatra is reported, and economic chaos as the scale of the events overwhelms the poor local government authorities, wiping out thousands of small businesses, civil infrastructure. This huge event is nevertheless limited for the most part on the margins of the contenent, but is unbounded for some of the island nations.
Entire Island populations may be gone, there are thousands of islands to check, few to do the checking.