Costa Rica evacuates at 10 a.m. today. The alarms are not real, you should go along with whoever asks you to leave the building, and you will be back at your desk in about fifteen minutes. Worth knowing while you are standing in the parking lot: the network that would detect the real thing has been running without dedicated funding since 2023.
The Observatorio Vulcanológico y Sismológico de Costa Rica, run out of the Universidad Nacional, operates more than 110 high-sensitivity seismic stations. Those stations are the reason your phone can tell you the magnitude and epicenter of a tremor within a minute of you feeling it, and they feed the early warning system this morning’s exercise is meant to prepare you for.
The money that kept them running came from a temporary provision of the National Emergency Law, drawn from the Fondo Nacional de Emergencias. It expired in 2023 and was never renewed. Esteban Chaves, who directs the observatory, has laid out the split plainly. The Universidad Nacional pays the salaries of the people who work there.
It does not pay to maintain the instruments in the field, replace the ones that fail, or expand the network. That was what the expired provision covered. The observatory has warned that stations dropping offline one at a time would bring back what it calls blind spots, sections of the country where nothing is recording at all. Equipment sitting outdoors through Costa Rican rain and humidity does not last indefinitely without maintenance.
What that costs you is easier to see in a specific event. On July 7 at 12:11 p.m., a magnitude 5.2 quake struck 28 kilometers (17 miles) southwest of Jacó at a depth of 22 kilometers (14 miles), on the interface where the Cocos plate slides beneath the Caribbean plate. It was felt across the Central Pacific and the Central Valley. Nobody was hurt and nothing structural failed.
By 1:40 that afternoon the observatory had logged 21 aftershocks. By late afternoon the count had reached 38, the largest at magnitude 3.1, and Chaves was able to say publicly that the sequence matched normal subduction behavior rather than the opening act of something worse.
That reassurance is a product. It exists because stations were recording. Costa Rica averages more than 3,000 tremors a year, the overwhelming majority of which nobody feels, and the difference between a quake that rattles people and a quake that starts a panic is usually somebody with data saying which one it is within the hour.
The stations also do slower work. In 2025 and 2026, observatory researchers identified an active system of shallow crustal faults responsible for tremors felt in the urban area of the capital, a finding possible only because instruments had been recording continuously for years beforehand.
A fix has been sitting in the Legislative Assembly. Bill 24.738, drafted by Liberación Nacional deputy Rosaura Méndez, would add an article to the National Emergency Law creating permanent financing for the country’s monitoring networks through a 0.60% contribution on insurance premiums.
The observatory would receive roughly ₡600 million a year, about $1.33 million at the Central Bank reference rate of ₡452.70 on August 11. The Red Sismológica Nacional and the seismic engineering laboratory at the Universidad de Costa Rica, along with the Instituto Meteorológico Nacional, would be funded the same way.
The Instituto Nacional de Seguros objected to it. Appearing before the Assembly’s social affairs committee, INS president Gabriela Chacón argued that a new charge on premiums could hurt insurance customers and destabilize the market, and said the country should not solve one shortfall by creating another.
The reference point for all of this is April 22, 1991, when a magnitude 7.7 quake with its epicenter in Valle de la Estrella killed 48 people, injured more than 650, and lifted sections of the Caribbean coastline by around 1.5 meters (5 feet). The observatory’s argument is that the country was blind then and should not be blind again.
The drill teaches you what to do in the ten minutes after the ground stops moving. Where to meet, which stairs to take, who collects the children. That part is worth doing, and it costs a quarter of an hour once a year.
The other half of the system is the part that tells us what just happened, how big it was, and whether more is coming. Nobody rehearses that one. It either works when the moment arrives or it does not, and right now it runs on a budget line that ran out three years ago.
