(CNN) -- The Indonesian government has defended itself against accusations it issued inadequate safety warnings as a powerful earthquake triggered a tsunami that killed at least 844 people.
The 7.5 magnitude earthquake hit Sulawesi island on Friday causing three-meter-high (10 foot) waves to surge over parts of its northwestern shore, with groups of people caught on the coastline, apparently unaware of the danger.
Experts underscored the difficulties of predicting tsunamis and raising safety alerts across Indonesia's vast archipelago of over 17,000 islands and 261 million people, but others pointed to a lack of government coordination and funding.
Online criticism from Indonesians centered on allegations that the country's meteorology and geophysics agency (BMKG) canceled the tsunami alert too early.
The alert was sounded shortly after 6 p.m. on Friday warning of potential waves of up to three meters. The warning was canceled at 6.36 p.m. However, the agency said the alert was lifted only after the tsunami hit.
In a statement, the head of the organization, Dwikorita Karnawati, called the allegations "not correct."
"Our work is based on computer system/artificial intelligence. The warning system was lifted with the approval of the 28 other countries along the Indian ocean," said Karnawati.
Phil Cummins, professor of natural hazards at Australian National University told CNN that while all the facts have yet to emerge, he doubted the warning sign issue was a major concern.
"During the time when a warning was potentially useful, it was apparently in effect, and that seems far more significant to me than having canceled it prematurely after the tsunami hit," he added.
Did the tidal buoys work?
Questions have focused on Indonesia's warning system, founded on a grid of 134 tide gauge stations and land seismographs, bolstered by education campaigns to train people in tsunami response.
Its warnings are transmitted by sirens and text messages.
"From the reports I've heard in the media, they (sirens) weren't operable because the power was knocked out by the earthquake," said Cummins. "If true, that's something that should really be looked into."
CNN has not been able to confirm these reports.
Experts also focused on the number of tidal gauge systems, which measure sea level at the coast, that are in use.
"134 tide gauge stations are not adequate to confirm tsunami generation for the country which perhaps has the most islands in the world," said Professor Costas Synolakis, director of the University of South California Tsunami Warning Center.
"To be effective, you need to have a tide gauge near every coastal community," he added.
Concerns over the government's readiness increased after Sutopo Purwo Nugroho, the spokesman for the country's national disaster agency (BNPB), admitted that none of the additional 22 deep water buoys connected to seafloor sensors to help monitor for tsunamis off Indonesia's coast had worked for the last six years.
The buoys are intended to work in addition to the tide gauge stations as part of Indonesia's Tsunami Early Warning System, designed after the devastating 2004 Aceh tsunami.
"Disaster funding continues to decline every year. The threat of a disaster increases, the incidence of disasters increases, the BNPB budget goes down," said Nugroho, according to a report by CNN Indonesia.
Shape of bay amplified waves
International experts said that while Indonesia's warning system might be improved, its geographic position and the nature of the disaster made tsunamis almost impossible to defend against.
"First we need to realize that each tsunami is unique and therefore very difficult to predict particularly where there are not long periods between the earthquake and tsunami," said Charitha Pattiaratchi, professor of oceanography at the University of Western Australia.
He added that the epicenter of the earthquake was on land, so had a very low probability of generating a tsunami, suggesting that it may have been triggered not by the earthquake itself but by a submarine landslide.
Pattiaratchi said that Palu, the largest city to be hit by the tsunami, is located at head of a long narrow bay, measuring around 30-40 km in length. "As the tsunami traveled up this bay it would have become highly amplified."
Adam Switzer, a tsunami expert at the Earth Observatory of Singapore, said that Indonesia's primary warning system -- a database model that runs various scenarios based on the early parameters of the earthquake-- "appears to have worked."
The earthquake sequence was complicated, he added, with multiple aftershocks and a very high likelihood of one or multiple landslides either above or below the ocean surface.
"This is a clearly a very unusual event and the satellite imagery and eyewitness videos are showing a number of unusual characteristics that will take months to work out," said Switzer.
"No tsunami warning system I know of can handle multiple shocks or landslides as it is just too physically and computationally complicated for current technology," he concluded.
Also, in Indonesia's case, its buoys and gauges are mostly positioned in areas considered more likely targets for tsunamis, which Sulawesi was not, said Pattiaratchi.
"This is not a region where large tsunamigenic earthquakes are reported to have occurred in the past," he added.
"With limited funds everyone have to take decisions to make priorities. So if there was no reason -- based on historic events -- then decisions are made."