“Photonic Sensing” is the somewhat highfalutin term for a family of emerging technologies that use lasers and fiber-optic cables as ground motion sensors, not just as telecommunication conduits. A recent article in UW News has brought the subject of photonic sensing to the attention of our fans. So, this blog post is a very quick and dirty primer on the promise and problems of fiberoptic sensing in regional seismic monitoring. The subject is vast, decidedly complicated, and rapidly evolving…so hang on to your hat and let’s go!

How photonic sensing works

Fiber-optic sensing, or photonic sensing, or distributed sensing all refer to methods based on firing light pulses from a high-powered laser down through a fiber-optic cable and using characteristics of either the transmitted—or reflected—light pulse to “interrogate” the state of strain within and along the fiber. That state of strain can further be related to:

• Stretching and contraction along the axis of the fiber – Distributed Strain Sensing (DSS);
• Vibration of the fiber along its axis (like a “seismometer”) – Distributed Acoustic Sensing (DAS); or
• Temperature along the fiber – Distributed Temperature Sensing (DTS).

The secret to photonic sensing is the “scattering” of the laser light by natural imperfections within the tiny fiber.

The most relevant technique for us seismologists is DAS, so let’s take a quick dive into how that works. The most fundamental description is that a laser light pulse is fired from an “interrogator unit” into one end of a pre-existing fiberoptic cable and energy scattered back to the interrogator is analyzed to determine the state of strain along the fiber (see Figure 1). The backscatter is due to imperfections in a fiber that scatter light in several physical ways. DAS relies on “Rayleigh scattering,” in which an imperfection sends a ripple of energy back towards the direction the laser light pulse came from. Each fiber has its own random pattern of imperfections that was cast when the fiber was fabricated. If the fiber is strained a bit (stretched or contracted) somewhere along it, the imperfections will be moved closer together or farther apart in that section just a tiny bit.

Figure 1. Schematic of fiber-optic DAS technology basics. The interrogator unit sends a pulse of laser light outbound (in this diagram it is at the end of the fiber and traveling away from the interrogator). As it passes through the fiber, imperfections (red dots) scatter some light back toward the interrogator, where the pattern is detected and analyzed. Strains in the seismic wavefield deform the fiber and perturb the pattern. The resulting data can be displayed as an array of broadband “seismograms”, shown at the lower left.

Let’s get back to the back-scattered energy, which will pop out of the fiber where the laser pulse was introduced, right? The DAS technique measures the changes in back-scattered intensity between successive pulses that come from many “channels” along the fiber. These changes in intensity are correlated with changes in the optical path length (strain) along the fiber. The technique of interferometry is used to extract any changes in the strain state along the fiber, and thus produce time series of ground vibration. I find it easier just to think of the backscattered energy as being a random “code” that is correlated between successive pulses. This code gets distorted—alternately stretched and compressed—by pressure waves that are in the medium within which the cable is contained (i.e., the ground!). Interferometry detects and identifies the stretching and compressing distortions of the code  By keeping track of the “time of flight” of the pulse the interrogator determines where along the fiber the strain is being measured. 

Now that’s amazing. But it’s just the start of amazingness. For while interferometry is an olde timey technique dating from the late 1800s, its use with lasers at extremely high sample rates controlling the lengths of light pulses within a fiber, and with the computing power to keep track of it all is really jaw-dropping. Because by manipulating the strength of the laser pulse, its duration in time, and how frequently you sample (i.e., send a light pulse down the cable), you can adjust how many sections your fiber can be divided into, and what frequencies of ground motion you can resolve. That’s right: your fiber potentially can have tens of thousands of channels (i.e., measurement points) along its length, spaced on the order of several meters. And because the laser can be fired at kilohertz rates, a single fiber can provide data equivalent to thousands of broadband seismometers. This is, obviously, potentially revolutionary, and it’s the reason the seismological community is agog. 

The potential uses of DAS data extend far beyond network seismology. Thus far, development of instruments and analytical methods has largely happened in the fields of geotechnical monitoring, intrusion monitoring, and industrial infrastructure state-of-health monitoring. But the new Photonic Sensing Facility at UW and its partnership with the PNSN aims to research how best to integrate DAS technology into the monitoring of regional seismic hazards. 

While DAS can use custom-built fibers for observations, it also (and perhaps more importantly for our purposes) can take advantage of pre-existing telecommunication fiber-optics – called “Dark Fiber” (because it’s not lit…get it?). And there is LOTS of dark fiber around. There are fibers under the city streets, there are fibers in cables between cities, and, perhaps most important for PNSN, there are dark fibers in the telecom cables that run offshore and girdle the Cascadia Subduction Zone. Hmmm…30,000 broadband seismometers (not to mention strain meters from DSS) already hanging out on top of our biggest seismic hazard. No wonder we’re excited (see Figure 2)!

Figure 2. Example of a subset of data from a recent DAS experiment on a dark fiber in an offshore cable across the Cascadia continental shelf. The green bars at the top summarize the total backscattered energy along the fiber. The interrogator was on the left side of the plot. The bottom shows 3 minutes of data for each of about 32,000 channels of the total of 40,000. Each vertical column is thus a 3-minute-long broadband seismogram depicted as a red-to-blue pattern—most recent data at the top— instead of the more common “wiggle”. Bright bands in the plot fingerprint different sources in the seismic wavefield. Steeply plunging bands at the left side are from ocean waves propagating toward the shoreline slowly. The series of sharp repeating flatter bright lines in the center left are the chirping vocalizations of fin whales.  And the light blue wash across the center right channels about a quarter of the way down the plot were contributed by the somewhat mysterious “T-phases” from an earthquake far away from the cable (near Cape Mendocino).

Challenges of DAS

To determine what developments are needed to make DAS useful in monitoring there are several clear challenges (which I think are all addressable), and things we just plain don’t know (That’s the fun part!).

The challenges include:

• Each virtual seismometer is most sensitive to only one direction of ground motion: the linear axis of the fiber. Regular seismometers detect motion in all three directions. It will be important to understand how best to use this 1-dimensional observation in our regular processing.
• Dark fiber is generally in cables and the cables weren’t necessarily installed such that they were uniformly “tied” to the ground.  So, the strains or vibrations we observe may record the real ground motion with varying fidelity along the fiber. This will take the development of techniques to quantitatively characterize the “transfer function” of each dark fiber, and perhaps each channel along a fiber.
• Particularly in urban areas, we have found that cables may have been placed in slightly different locations than the current owners think they were. Surprise! There are also often loops of fiber (called “service loops”) every so often along a fiber that are there for any needed repair. This “extra” fiber length means your channels geographic distances (i.e., the mapped locations) may be shorter than the distance the laser pulse traverses (the “optical distance”). To unravel these complications, it is often necessary to perform “tap tests” (so called because the usual source is a student “tapping” the ground with a sledgehammer near the fiber at known locations and times) to accurately place fiber channels on a map.
• DAS produces vast, huge, overwhelming amounts of data. A years’ worth of seismic data from all of the current stations of the PNSN network could easily be surpassed by a single DAS experiment on a 60 km length of fiber in a couple of weeks’ time. Of course, we may not need to archive all that data as we currently do with the PNSN network data, but we will need to decide what data to keep and build a way to store it.
• Speaking of Big Data, machine learning and artificial intelligence will undoubtedly be required to sort for patterns in DAS data, perhaps finding the “needle in the haystack” of important signals by de-noising the data or using other deep learning techniques to recognize important “events” in the firehose of new seismic data exploding through our system.
• DAS data are, for many reasons, not really the same as PNSN data that all use standard inertial seismometers of one sort or another. So, the “metadata,” or how those data are described and organized, must be adapted so that the data are useful in our analysis programs and can be exchanged with other networks and researchers.
• And finally, how will regional networks like PNSN use DAS data in operation? Earthquake early warning from offshore fibers seems like an obvious use. Characterizing ground motion, and possible site effects, on a building-by-building scale may be possible. Using data from fibers that simply enable us to extend the number and locations we can monitor; sure. But these may be only the tip of the iceberg.

This is truly an exciting, energizing, and revolutionary time to be an observational Earth scientist. And the PNSN and the UW Photonic Sensing Facility are proud to be at the point of the DAS spear!