🔬 physics

Method for restoring the orientation of ocean bottom seismometers using distant earthquakes records

This paper presents and validates a method for restoring the orientation of ocean bottom seismometers by leveraging the linear relationship between vertical seafloor acceleration and bottom pressure variations during distant earthquakes, demonstrating its effectiveness across six S-net stations with high accuracy and applicability to both accelerometers and velocimeters.

Original authors: Oleg V. Ponomarev, Sergey V. Kolesov, Michail A. Nosov

Published 2026-06-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Oleg V. Ponomarev, Sergey V. Kolesov, Michail A. Nosov

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the ocean floor is a giant, dark stage where scientists have placed sensitive microphones (seismometers) and pressure sensors (like underwater barometers) to listen to the Earth's rumblings. But there's a catch: when these heavy instruments sink to the bottom, they don't always land perfectly upright. They might tilt, twist, or lean like a drunk sailor on a rocking boat. If the "microphone" is leaning, it hears the Earth's voice wrong, mixing up up-and-down movements with side-to-side shuffles.

This paper presents a clever, self-correcting trick to figure out exactly which way is "up" for these tilted instruments, without needing to send a diver down to fix them.

The Core Idea: The "Water Elevator" Analogy

The scientists rely on a simple rule of physics, similar to how an elevator works. Imagine you are standing in a glass elevator filled with water.

If the elevator accelerates up, the water pushes down harder on the floor (pressure goes up).
If the elevator accelerates down, the water pushes less hard (pressure goes down).

There is a direct, predictable link between how fast the floor moves up and down and how much the water pressure changes. The paper calls this the "forced oscillation" range—a specific frequency of shaking where the water acts like a solid block moving with the floor, rather than sloshing around like a wave.

The researchers realized: If we know how much the water pressure changed, and we know how fast the floor moved, we can check if our sensors are telling the truth.

The Method: The "Shadow" Hunt

Here is how they fixed the orientation of the sensors using distant earthquakes:

The Setup: They looked at data from six underwater stations (part of a network called S-net) during three massive earthquakes (in Alaska, Chignik, and near Japan). These quakes were strong enough to make the ocean floor shake in that specific "elevator" frequency range.
The Guessing Game: The computer took the raw data from the tilted sensors (which are labeled X, Y, and Z) and started playing a game of "What if?"
- It imagined tilting the sensors in every possible direction (like rotating a globe).
- For every single angle, it calculated a "shadow" of the movement.
The Match: For each imagined angle, it asked: "Does the 'shadow' of this movement match the pressure changes in the water?"
- If the sensors were perfectly aligned with the vertical direction, the movement shadow and the water pressure would be a perfect match (like two dancers moving in perfect sync).
- If the sensors were tilted, the match would be messy and out of sync.
The Result: The computer found the specific angle where the match was the strongest. That angle told them exactly which way was vertical relative to the instrument.

The "Two Faces" Problem

There is one small quirk in this method. Because the physics works the same whether you are pushing up or pulling down, the method can't tell the difference between "straight up" and "straight down." It finds a line, but it doesn't know which end is the ceiling and which is the floor. However, for the purpose of knowing how the instrument is tilted, knowing the line is usually enough.

What They Found

It Works: When they compared their new "water pressure" method with an old method that just looks at the constant pull of gravity (like a level on a carpenter's tool), the results were almost identical. For most stations, the difference was less than 1 degree—about the width of a pencil line.
It's Robust: They tested this on three different earthquakes over several years. The instruments didn't seem to shift their tilt over time, which is good news for scientists relying on them.
The Deep Water Problem: One station was very deep (6km). At that depth, the "elevator" effect requires a very slow, deep rumble to work. The earthquakes they used weren't quite strong enough to make the water behave perfectly at that depth, so the "match" was fuzzy. It's like trying to tune a radio station that is too far away; the signal is there, but it's staticky.
The "Smeared" Signal: During the earthquake closest to the sensors (the Noto quake), the "perfect match" lines looked a bit blurry. The scientists think this is because the shaking was so violent that the instruments actually wobbled or shifted slightly during the event, like a camera shaking while taking a photo.

Why This Matters

The biggest win of this paper isn't just fixing the sensors; it's that this method works for both accelerometers (which measure shaking speed) and velocimeters (which measure shaking speed). Previous methods often only worked for one type. Now, scientists have a universal "compass" to ensure their underwater ears are listening in the right direction, no matter what kind of instrument they are using.

Technical Summary: Method for Restoring the Orientation of Ocean Bottom Seismometers Using Distant Earthquake Records

Problem Statement
Ocean bottom seismometers (OBS) installed in seafloor observatories, such as the S-net network, often suffer from misalignment of their sensor axes relative to the true vertical direction. This misalignment can result from installation errors or shifts caused by strong seismic motions. While existing methods exist to recover accelerometer orientation based on the constant gravitational component (Takagi et al., 2019), these approaches are not universal; they cannot be applied to velocity sensors (velocimeters). Furthermore, strong nearby earthquakes can alter sensor axes, necessitating robust methods to determine the vertical direction relative to the instrument's coordinate system.

Methodology
The proposed method determines the vertical direction relative to seismometer axes by exploiting the linear physical relationship between seafloor vertical acceleration ( $a_v$ ) and ocean bottom pressure variations ( $p$ ). This relationship, derived from Newton's second law, holds within the frequency range of "forced oscillations" ( $f_g < f < f_{ac}$ ), where $f_g$ and $f_{ac}$ are characteristic frequencies dependent on water depth ( $H$ ) and sound speed ( $c$ ). The governing equation is:
$p = \rho H \cdot a_v$
where $\rho$ is the vertically averaged water density.

The algorithm proceeds in three stages using cross-spectral analysis:

Data Preparation: Pressure data is detrended to isolate dynamic variations. Acceleration data from the three orthogonal axes ( $x, y, z$ ) is resampled to match the pressure sampling rate.
Projection and Coherence Calculation: The method generates a set of acceleration projections ( $a_{\eta,\kappa}$ ) along arbitrary spatial directions defined by angles $\eta$ (elevation from the $z$ -axis) and $\kappa$ (azimuth in the $xy$-plane). For each direction, the Magnitude-Squared Coherence (MSC) between the pressure variation and the projected acceleration is computed.
Optimization: A coherence integral ( $I$ ) is calculated over the valid frequency range ( $f_g$ to $\min(f_{ac}, 0.1\text{Hz})$ ). The direction corresponding to the maximum integral value ( $I_{max}$ ) is identified as the vertical axis. The method utilizes Welch's method for spectral estimation.

Key Contributions

Universal Applicability: Unlike gravity-based methods, this approach is applicable to both accelerometers and velocimeters, as it relies on dynamic signal coherence rather than static gravity components.
Validation Framework: The method was rigorously tested using data from six S-net stations during three distinct seismic events: the Gulf of Alaska $M_w7.9$ (2018), Chignik $M_w8.2$ (2021), and Noto $M_w7.5$ (2024).
Depth-Dependent Analysis: The study explicitly addresses the limitations imposed by water depth, noting that the forced oscillation frequency range shifts to lower frequencies as depth increases, requiring larger magnitude earthquakes to effectively excite the necessary signal band.

Results

Accuracy: For four of the six tested stations (located at depths $\approx 2$ km), the orientation angles ( $\eta, \kappa$ ) derived from this method differed by less than $1^\circ$ from those obtained via the independent gravity-based method.
Consistency: No significant changes in sensor orientation were observed across the six stations between 2018 and 2024.
Depth Limitations: Station S6N08, located at a depth of 6 km, exhibited less distinct coherence peaks and larger discrepancies in calculated angles. The authors attribute this to the fact that the analyzed earthquakes ( $M_w 7.5–8.2$ ) were insufficient to excite the full range of forced oscillations required at such depths (where the lower frequency bound is $\approx 0.015$ Hz).
Signal Intensity Effects: For the Noto earthquake, which occurred close to the network, coherence diagrams for some stations appeared "smeared." This suggests dynamic changes in the station's spatial orientation during the passage of strong seismic motions.
Ambiguity: The method determines the vertical direction only up to a sign (i.e., it identifies the axis line but cannot distinguish between "up" and "down" without additional assumptions).

Significance
The paper establishes a reliable, physics-based method for calibrating the vertical orientation of ocean bottom sensors using distant earthquake records. Its primary significance lies in its ability to validate sensor orientation for velocimeters, a capability lacking in previous gravity-based techniques. The study confirms that for stations in moderate depths (approx. 2 km), distant large-magnitude earthquakes provide sufficient signal to determine orientation with high precision ( $\le 1^\circ$ ). The authors note that the method's effectiveness is constrained by water depth and earthquake magnitude, as deeper stations require lower-frequency excitation that may not be fully captured by the analyzed seismic events.