On the topic of Earthquake Early Warning for Puerto Rico

I write this article mostly to have something to refer people to when they bring up the topic of an earthquake early warning (also known as EEW) in the Puerto Rico area. I do it to mostly answer the question they frequently ask me.

Is it possible to have a system that tells me when there will be an earthquake before the ground starts shaking?

No*

It’s that easy to answer, but I don’t just want to answer it and move on, here I will present some arguments that can be made to explain why it is simply not possible on the scale that one would like, or that would make sense to use for non-special users. Understand that “no” has an asterisk for a reason. We will begin to explain with the most basic argument: How do we detect earthquakes? How do we determine where their origin is? Assume for a moment that an earthquake occurs:

This is an animation of the p-wave propagation of a possible earthquake in Puerto Rico of undetermined location for the first 30s. In the red circle, the primary wave is displayed, which is the first wave in which detections are made and with which information about the earthquake can be extracted. The speed of the wave is fixed at 7000 m/s or 7 Km/s, something that is typical for the earth’s crust.

Using 3 or more stations, a network can find a location for an event, the more stations the better the location. In the case of wanting an early warning, we will assume the best case scenario:

We have stations very close to the earthquake area (within 2 s), this is not always the case, but we are going to make an assumption.
They are connected through a reliable communication system with low latency, the information will be in the system in 2 s
The system in which it is connected is superfast, efficient, 100% automated and infallible, it will take about 30-50s to make an earthquake determination.

These are the 3 assumptions we are going to make to run this simulation. What would this look like?

This figure shows a possible sample of stations (yellow triangles) that are theoretically within 2 s of the earthquake. With these stations we can begin to generate a solution that tells us the location and magnitude of the earthquake.

Excellent, the 3 stations will have captured the earthquake and are sending the information, this is excellent news, but it will be another 2 seconds until the data center that processes this information arrives. After that, it will be another 40 seconds before we know where the earthquake is and its possible magnitude. Let’s see where the wave front is in 44 s:

Figure shows the simulation of p-waves at 40 s.

Ohh…. Ok… well then.

How early is “Early Warning”?

Well, one thing that will help the case is that the p-wave front is not the front that causes damage. Secondary waves (s-waves) and surface waves are those that cause damage to structures, which, luckily, are significantly slower! Let’s run the simulation again, luckily we already know where the earthquake marked with a star was:

This is an animation of the p-wave and s-wave propagation of a possible earthquake in Puerto Rico of undetermined location for the first 50s. In the red circle, the primary wave is displayed, which is the first wave in which detections are made and with which information about the earthquake can be extracted. In the blue circle the secondary wave is displayed, which is the precursor to the surface waves, the s-wave and the surface waves are the ones that create the damage in an earthquake. The speed of the primary wave is fixed at 7000m/s or 7 km/s which is typical for the earth’s crust, and the secondary wave is fixed at 3300m/s or 3.3 km/s which is average for the earth’s crust.

Here then something a little more promising, in which for 44 s at least it gives time to the people on the island of Culebras and the eastern part of Vieques to receive the alert that would really take 1-2 s to spread from one data center to the rest of the island:

Simulation of p and s waves for 45 seconds.

This gives us a better intuition of how an earthquake behaves. Given these differences in speed one can easily visualize that the further one is from the earthquake, the longer it takes for the wave front that causes damage to arrive, this is the time one needs to do what is necessary to prepare and issue an alert. If the earthquake is far enough away, a system can detect, locate and issue an earthquake alert, before you feel any shaking.

Can the system be improved?

Let’s look at other examples around the world, the most famous being Japan, whose system also exists on an island and is much more important given its distributed system of natural gas and high-speed trains:

Figure shows on the left what the alert can be used for and which stations are used. On the right a figure of the seismometers and ocean bottom seismometers. Retrieved from https://www.jreast.co.jp/e/development/theme/safety/safety07.html on March 12, 2024.

Okay, in that figure you can see that not only do they use more seismometers, but the seismometers can be submerged under the ocean to be able to detect the presence of an earthquake much earlier, almost immediately. So in this way we can try to improve an early warning system in Puerto Rico by placing sensors where there may be earthquakes. However, for a moment, let’s also take into consideration the distances involved:

Figure shows the area of Japan. A star marks the location of the 2011 Tohoku earthquake. Superimposed on it is the silhouette of the central island of Puerto Rico (in green) for comparison of the distances involved.

Okay! Now, in this new context, you can see why an early warning system for Japan works. A combination of detectors above the earthquake zone, even below the ocean, with distances that are much greater. The earthquake occurred almost one whole Puerto Rico away. You can imagine that running the simulation, there is much more time between detection and alert to a metro area like Tokyo or Osaka, with areas closer to the earthquake like Fukushima barely receiving the alert with the waves. Assuming of course an extremely fast and accurate system, it would be enough to be able to do high priority things, for trains, gas, and electricity high speed shut-offs! Will such a system be possible in Puerto Rico?

Who is the system for?

Japan’s system and other similar systems exist, largely, not to inform the population, but rather go hand in hand with critical systems to be able to manage a crisis before it begins. In Japan the trains stop quickly, those that are not moving stay still. High pressure gas lines are closed with automatic valves. Generation systems begin to disconnect high voltage lines from the electrical grid. In short, many critical things begin to prepare for possible complications following an earthquake. The most important clients of such a system are commercial or technical in nature.

In addition, those areas that have not begun shaking may receive an alert for people to seek safety for when surface waves arrive. The only for areas in which this is possible, which can be defined with a distance at which the system has had time to detect and locate the earthquake, and the alert issued could have been sent with enough speed to beat the surface waves to the people. The speed at which you can run many of the steps needed is important, but the lower limit will always be physics! Think about:

Detection of the instrument by location.
The speed of information transmission by radio, cell phone, fiber optics, etc.
The speed and capacity of the computer doing the calculations.
The speed of verifying that it is an earthquake, even more so if someone has to check it.
The speed of alert information emitted by radio, cell phone, fiber optics, email, etc.

All of this can delay the system! For a place where it is of utmost importance like Japan, they have spent a lot of time and money developing it (including putting instruments underwater). More than that, they have distance! Long enough to delay the waves to the point where the speed of the alerts can overtake the speed of the earthquake itself. The alert arrives before the waves reach the most populated locations and high-priority equipment. For many places, these vital stations can be connected with fiber optic lines, indicating that they communicate at the physical speed of light, in addition the computers doing the calculations and locating are both distributed, and among the fastest and most sophisticated. Their alert system is integrated with the cell phone, television and radio systems. It is a system diametrically different from others. In the western area of the US there are similar systems, but in Puerto Rico it is a different story…

Observations and Conclusion

So what is “Early” for an early warning system? It depends on many factors:

The location of the earthquake.
The location and distribution of seismometers.
The speed at which they can transmit the data to the collection center.
The speed at which a system can accurately differentiate false triggers from earthquakes and automatically produce an alert.
The speed at which the alert can spread.

Finally, for earthquakes near or within Puerto Rico, an early warning system is neither easy nor trivial, if not entirely physically impossible. I have been, in my opinion, very generous in the times and speeds described here. So where are we left with the “No*”?

Should it exists, it would be used for earthquakes that are a little further away, or it would need a small, local, and low latency networks. It would not be an alert for us human beings, but for services such as the urban trains, electric stations, and gas lines. They are the only systems that, given the correct conditions, can, in less time than it takes us to open and close an eye, be able to close gas valves, turn off and protect electricity systems, and stop a train on time. That is sort of describing existing systems, or systems that should already exist. Systems that only have detection in the area of the train, or the electric generation plant, or in critical places. They would be systems that the public does not see, but exist to maintain security.

The best early warning system is the human being himself and being educated on what to do before, during, and after an earthquake. Perhaps, for other things, early warning systems work very well, for example: tsunamis take minutes to arrive, hurricanes days to arrive. Until technology improves, it’s best to simply be more alert and educated about earthquakes.