🔬 physics

An example of regional seismicity recovery on June 22, 2020, using waveform cross-correlation

This paper investigates the seismic activity of June 22, 2020, a date marked by conflicting claims regarding a potential nuclear test at Lop Nor, by applying advanced waveform cross-correlation techniques to re-examine data and determine if a hidden event occurred below the standard detection threshold of the international monitoring system.

Original authors: Ivan Kitov

Published 2026-03-03

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Ivan Kitov

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

The Big Mystery: The "Ghost" Explosion

Imagine a world where countries have promised not to build nuclear bombs. To keep this promise, there is a global security system (the CTBTO) that acts like a giant, super-sensitive set of ears, listening to the Earth for any underground "thumps" that might be secret tests.

On June 22, 2020, a US official claimed a secret nuclear test happened in a specific desert in China (Lop Nor). However, the global security system said, "We heard nothing."

This created a mystery:

Did the test happen but was too quiet to hear?
Did the security system miss it?
Or did nothing happen at all?

This paper is an investigation to see if the "ears" were just not listening closely enough.

The Problem: The "Standard Ear" vs. The "Super Ear"

The global security system uses a standard method to listen for earthquakes and explosions. Think of this method like a standard microphone in a noisy room. It can hear someone shouting (a big earthquake), but if someone whispers (a small, hidden explosion), the microphone misses it because the background noise (wind, traffic, distant rumbling) drowns it out.

The authors of this paper wanted to see if they could build a "Super Ear" using a technique called Waveform Cross-Correlation (WCC).

The Solution: The "Song Match" Analogy

How does this "Super Ear" work? Imagine you are trying to find a specific song playing in a crowded, noisy club.

The Standard Method: You just listen for any loud noise. If the music is too quiet, you miss it.
The WCC Method: You have a recording of the exact song you are looking for (a "template"). You play that recording in your head and listen to the club. Even if the song is playing very quietly, your brain recognizes the pattern of the melody matching the background noise. You can pick out that specific song even if it's 10 times quieter than the noise around it.

In this paper, the "song" is the seismic wave from a known earthquake. The scientists used past earthquakes as templates to scan the data from June 22, 2020, looking for that same "melody" hidden in the noise.

The Investigation: What Did They Find?

The scientists took the data from June 21, 22, and 23, 2020, and ran it through their "Super Ear" (the WCC method).

The Results:

They found 30 new "events" that the standard system missed.
Most were industrial: The timing of these events (mostly during the day) suggested they were caused by mining or blasting rocks, not nuclear bombs. It's like hearing a construction crew drilling in the distance; it sounds like a thump, but it's just work.
The "Ghost" Test: They did not find a hidden nuclear explosion on June 22. The only real event the standard system found was a small, natural earthquake (magnitude 3.87).

The "Magic Cave" Concept

The paper also discusses a clever trick used to hide nuclear tests called Cavity Decoupling.

The Analogy: Imagine you want to scream in a cave without anyone outside hearing you. If you scream in a small, tight tunnel, the sound bounces around and gets loud. But if you scream inside a giant, hollow, spherical ballroom, the sound waves hit the walls and cancel each other out.
The Science: If a country detonates a bomb inside a giant, hollow underground cave (about 20 meters wide), the seismic "thump" can be reduced by a factor of 100. It turns a "shout" into a "whisper."
The Catch: Even if the sound is hidden, the explosion might melt the rock and cause cracks, letting dust or gas leak to the surface. So, while the sound might be hidden, the mess might give it away.

The Conclusion

The paper concludes that:

The "Super Ear" works: The Waveform Cross-Correlation method is much better at finding small events than the standard system. It found many small mining blasts that the standard system ignored.
No Nuclear Test Found: Despite the US claim, the scientists found no evidence of a nuclear test on June 22, 2020. The events they found were likely just mining operations.
The System Needs Upgrades: The standard global monitoring system is good, but it misses small events. By using these "Super Ear" techniques, we can lower the threshold of detection and make it much harder for anyone to hide a secret test.

In short: The scientists used a high-tech "pattern matching" trick to listen closer to the Earth. They found that while the Earth was indeed "talking" (with mining blasts), there was no secret nuclear whisper on that specific day.

1. Problem Statement

The paper addresses a specific challenge in nuclear test monitoring: the discrepancy between a public claim and official monitoring data regarding a potential secret nuclear test.

The Incident: On June 22, 2020, US Undersecretary Thomas DiNanno claimed a secret nuclear test occurred at the Lop Nor testing site. Conversely, the Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO) Provisional Technical Secretariat (PTS) reported no seismic events consistent with a nuclear explosion on that date.
The Gap: This discrepancy suggests one of three scenarios:
1. The event was hidden below the detection threshold of the International Monitoring System (IMS).
2. The event was missed by the International Data Centre (IDC) due to processing limitations.
3. No event occurred.
Technical Limitation: The standard IDC detection methods (primarily STA/LTA detectors) have a non-uniform global detection threshold. In continental areas, this threshold is often too high to detect weapon-sized explosions if they utilize cavity decoupling (burying an explosion in a large underground cavity to dampen seismic waves). The standard catalog (Reviewed Event Bulletin or REB) is known to be incomplete for magnitudes below $m_b \approx 4.0$ .

2. Methodology

The study employs Waveform Cross-Correlation (WCC) techniques to lower the detection threshold and re-examine the seismic data for the specified timeframe and region.

Data Scope:
- Timeframe: June 21–23, 2020 (to account for time zone uncertainties regarding the claim).
- Region: 38°N–45°N, 84°E–93°E (covering the Lop Nor area).
- Source Data: Continuous seismic data from the IMS network, including both primary and auxiliary stations (array and 3-component stations).
Master Event Selection:
- A global grid of nodes (1.5° spacing) was used to select "Master Events" from the existing REB.
- The selection criteria prioritized events with the highest "event weight" (a statistical measure of detection reliability) within a 0.75° radius of a grid node.
- A two-pass search was conducted: the first pass selected the best masters, and a second pass excluded those winners to find the next best candidates, resulting in a set of 76 master events.
Detection Algorithm (WCC):
- Template Matching: The WCC detector compares continuous seismic waveforms against templates derived from the selected Master Events.
- Correlation Logic: It searches for signals that are highly similar to the template. The method relies on the principle that events close in space and time will generate waveforms with high correlation coefficients at regional distances.
- Noise Handling: The study acknowledges that real ambient noise is not perfectly stochastic. However, WCC can detect signals with amplitudes as low as ~1/10th of the noise floor if the signal shape matches the template.
Phase Association (Local Association - LA):
- Once potential arrivals are detected, a Local Association procedure groups them into event hypotheses.
- Weighting: Stations are assigned weights based on their historical association frequency with events in that specific area.
- Thresholds: An event hypothesis is accepted if the sum of associated station weights exceeds a statistical threshold. Travel time residuals are constrained to 2 seconds (matching IDC REB standards).
- Conflict Resolution: If multiple master events compete for the same physical arrival, the hypothesis with the higher weight wins; ties are broken by the number of associated phases or RMS travel time residuals.

3. Key Contributions

Methodological Demonstration: The paper demonstrates the efficacy of WCC in recovering seismic events that fall below the standard IDC detection threshold, specifically in regions of high background seismicity and industrial activity.
Re-evaluation of the June 22 Claim: It provides a rigorous, data-driven re-analysis of the specific date in question, moving beyond the standard IDC pipeline to test the "missed event" hypothesis.
Refinement of Detection Limits: The study highlights that the standard REB catalog is incomplete for magnitudes below $m_b=4.0$ and that WCC can recover a significant portion of these "missing" events, effectively lowering the detection threshold by an order of magnitude in favorable conditions.

4. Results

Event Recovery: The WCC processing generated an XSEL (Cross-Correlation Standard Event List) bulletin containing 30 events across the three-day period:
- June 21: 12 events
- June 22: 12 events
- June 23: 6 events
Nature of Events:
- Timing Analysis: The distribution of events suggests a correlation with industrial activity rather than a single nuclear test. The majority (20 events) occurred between 06:00 and 15:00 UTC, a timeframe typical for open-pit mining and industrial operations. Only one event occurred between 00:00 and 06:00 UTC.
- Location: The detected events clustered in areas of known historical seismic activity (both natural and industrial), specifically near the north-western and south-eastern corners of the study area.
The June 22 REB Event: The only event officially recorded in the REB on June 22 occurred at 10:38 UTC ( $m_b=3.87$ ). The WCC analysis successfully identified this event but noted that the XSEL solution (automated location) was slightly offset from the REB location, suggesting the need for a better master event for precise localization.
Conclusion on the Claim: The study found no evidence of a hidden nuclear test on June 22, 2020. The 12 events detected by WCC on that day were consistent with the background seismicity and industrial patterns observed on the surrounding days.

5. Significance

Verification Capability: The paper validates WCC as a critical tool for the CTBTO and the broader scientific community to verify claims of clandestine nuclear testing. It proves that while standard methods might miss small events, advanced correlation techniques can push the detection limit lower.
Cavity Decoupling Context: The discussion contextualizes the results within the physics of cavity decoupling. While a 1kt explosion could theoretically be hidden in a ~20m radius cavity, the study implies that even if such an event occurred, the specific seismic signature (or lack thereof) on June 22 did not match the profile of a hidden test.
Future Monitoring: The authors suggest that WCC methods should be routinely applied to areas with salt deposits (where cavity decoupling is most effective) to ensure that even highly suppressed signals are not missed.
Scientific Integrity: By reprocessing the data with a more sensitive algorithm, the study reinforces the conclusion that the June 22 event was likely a false alarm or misinterpretation, thereby upholding the integrity of the CTBTO monitoring system while demonstrating how to improve it.