GPS Latency and EM Data: A Real-World Lesson in Time Shift Errors

Several years ago, a client shared EM data from a UST (underground storage tank) survey that identified three anomalies. Following the advice of a nationwide utility mark-out firm that had conducted the survey, she excavated all three locations—only to find nothing. Frustrated and losing faith in both the firm and possibly geophysics itself, she asked me how that could happen.

I explained that inaccurate GPS timing or odometer calibration could easily cause such ghost anomalies. I even sent her a simulation video narrating and showing how data can appear shifted due to latency or lag.

Fast forward to just a few weeks ago: Kyle Armendariz and I encountered a real-world example in our own dataset. During a nighttime survey within a street closure, we were configuring our EM61 (a time-domain electromagnetic instrument) and overlooked that it had defaulted to its internal GPS rather than the external receiver we intended to use. The result? A textbook case of GPS latency—and an extreme version of the dreaded herringbone effect.

Most of the time, when data isn’t corrected for latency—or when the odometer isn’t properly calibrated—you’ll see “wavy” anomalies. Linear features like pipes will appear to jog back and forth across the data. Our team is trained to recognize and catch these issues during field processing.

In this case, however, the problem was more extreme. The internal GPS is physically located behind the transmitter and receiver coils, introducing a negative latency. Unfortunately, the instrument’s software doesn’t account for—or allow—negative time shift corrections, which made the issue harder to identify at first. If we had intended to

What we knew to be a single paved-over trench plate appeared as two distinct EM anomalies. Neither aligned with the responses detected with the GPR or magnetic data, and both were far enough off to send a crew digging in the wrong places.

This is exactly how a client ends up digging two or more holes and finding nothing.

In the image below, the red “hot spots” (outlined in black polygons) are shifted far enough to miss the actual target—the brown square representing the paved-over trench plate. While we were initially scratching our heads, we shared the data with the client in the field and noted that we would evaluate the anomalies in real time. Each anomaly was marked on the ground and mapped using GPS. At the time, we suspected that we had selected an incorrect coordinate system or used imperial feet vs US survey feet. However, during final report preparation, we realized the extra anomalies were artifacts caused by a significant time shift error.

The survey passes—like a snake slithering across the site—dragged the anomalies back and forth, and offset from their actual location. Once corrected, everything snapped into place:

• The trench plate resolved as one coherent feature.
• The manholes lined up with their actual mapped locations.
• The phantom anomalies vanished.
• The north to south utility line became a clean, linear feature.

Takeaway: Always verify your GPS source and check for latency issues before delivering data. If using an odometer, calibrate it (or at least confirm calibration) before collecting data. A few seconds of lag—or a slight calibration error—can turn solid data into misleading targets.

Anyone else run into this kind of ghosting from GPS mismatch or timing issues? I’d love to hear your stories.