Yellowstone Caldera Chronicles is a weekly column written by scientists and collaborators of the Yellowstone Volcano Observatory. This week's contribution is from Scott K. Johnson, Science Communication Associate at EarthScope Consortium.

Sources/Usage: Public Domain. View Media Details

Given the level of interest in Yellowstone, it’s not uncommon to see people looking very closely at monitoring data. But without a deep understanding of the instruments and data processing techniques, it can be challenging to understand the causes of some signals—are you seeing some indication of geological activity, or an artifact or error in a measurement?

Take GPS data, for example, used to track ground motion in Yellowstone. The data plots you see on the YVO website and in the monthly update videos show daily average positions, but even after averaging a day’s worth of data, you can end up with an outlier. To the untrained eye, that might look like the ground suddenly moved that day—until normal-looking data in the following days demonstrate that, in reality, nothing happened.

Beyond the daily position numbers themselves, there are a number of checks that scientists and engineers use to evaluate data quality. The instrument logs the number of navigation satellites used for each measurement, and because fewer satellites means reduced accuracy, that might indicate a compromised data point. And because things like atmospheric conditions or objects partially blocking the signal introduce error, an important quality check is the variability of measurements throughout that day. Consistency is a sign of good data quality, while variability—indicated by large error bars on the daily position—is a good indicator of interference or other issues. “Filtered” data plots often exclude data points that fail these quality checks.

GPS data, which can measure changes of less than 1 centimeter (less than half an inch), are somewhat unique in that the initial values are often corrected after a few days, resulting in small changes that can reduce the datapoint-to-datapoint variability. This is because of after-the-fact updates to precise satellite tracking and atmospheric data, for example. These corrections are generally quite small, but they allow for more confident measurement of changes in the station’s position.

Sources/Usage: Public Domain. View Media Details

So what causes outlier days that stand out from the rest of the data once these variables are accounted for? All kinds of things—and nature constantly surprises us with new problems! For starters, any physical object between the GPS antenna and a satellite can bend or bounce the signal, slightly delaying its arrival at the antenna and affecting the position calculation. This is why these stations are placed in open areas with an unobstructed view of the sky. But if the nearest trees grow too tall over the years—or birds decide the antenna is their new favorite roost—the data quality can decline.

Most commonly, wintertime snow and ice accumulation can have the same effect. An ice storm can cause an apparent change in position that lasts for days—until ice melts off the antenna. Similarly, mountainous locations with large snowfalls often see increased variability in the winter as snow encroaches up to (or over the top of) the antenna. There generally is not a camera to give a visual of the station, and remote stations can be inaccessible in these conditions, but station history and weather reports can help us interpret patterns in the incoming data.

Another frequent culprit is malfunctioning equipment. This takes a multitude of forms, from electronics damaged by a nearby lightning strike to toothmarks made by overly friendly wildlife. Field engineers have to be prepared to troubleshoot and replace any part of the station when they visit. Often, that component is simply dead, but subtler problems like a loose or corroded cable can result in noisier data rather than missing data.

Permanent GPS stations are designed to minimize all these issues, withstanding the elements and turning satellite signals into the highest-quality ground position data possible. But especially in the harshest and most remote environments, the conditions eventually win. While stations are monitored for signs that they need a maintenance visit, remote sites aren’t always immediately accessible, so it can take time to make the repair.

Scientists know all this and are accustomed to evaluating data quality before drawing conclusions—sometimes discovering that a puzzling pattern is truly telling us something interesting about Yellowstone! So if you ever see a single data point circled by an excitable armchair expert online, remember to think like a scientist! Look for consistent trends rather than peculiar outliers and you won’t be fooled by a snowstorm or some tired birds.

Sources/Usage: Public Domain. View Media Details