Of 38 earthquake records in a selected southeastern Indian Ocean sequence, 37 give a depth of exactly 10 kilometres. The remaining record gives 9 kilometres. That repeated number is striking on a chart, so I followed the catalogue entries back to their individual origin records before interpreting it as a physical pattern.
The official USGS origin metadata answers part of the question: all 37 records at 10 km say the depth was “operator assigned.” The 9 km record describes depth estimation from modelling broad-band P waveforms. These are different descriptions of how a catalogue value was obtained. They do not show that 37 earthquakes were independently measured at precisely the same depth.
Reading the depth field's provenance
I compared all 38 selected QuakeML origin records with the frozen catalogue snapshot. An origin is a record of an estimated earthquake location and time. QuakeML is the structured format behind the displayed event: it can identify an origin's depth, the method description and uncertainty separately.
| Catalogue depth | Records | Origin's depth-type description |
|---|---|---|
| 10 km | 37 | Operator assigned |
| 9 km | 1 | Modelling of broad-band P waveforms |
All 38 origins also supply a depth uncertainty, ranging from 1.722 to 1.962 km, with metadata specifying 90% confidence. The uncertainty entries do not independently validate those depths or the models behind them. The retrieved origins contain no separate Boolean declaration that depth was fixed, so I have not relabelled the 37 assigned values as “37 confirmed fixed depths.”
Sixteen event-detail JSON responses were available. For the other 22, an official FDSN XML response identified the preferred origin, which led to its original QuakeML record. The selected product-version timestamps match the frozen catalogue's updated value for all 38 events. The current event-level updated value was directly checked for 16; it remains unknown for the 22 XML recoveries. Depth-method provenance was recovered for every event without downloading waveforms.
The empty historical circle depends on its radius
The catalogue comparison is centred on the first event's position, 47.8272° south and 115.9955° east. The recent query covers 7 June 2025 to 28 July 2026; the earlier query covers 1 January 1973 to 7 June 2025. All four query boundaries are at 00:00 UTC. Those unequal windows were chosen around a sequence already observed.
Using the same catalogue magnitude threshold, M ≥ 4.5, changes the historical comparison:
| Query radius | Earlier events | Recent events |
|---|---|---|
| 100 km | 0 | 24 |
| 200 km | 15 | 24 |
The same 24 recent event IDs appear at both radii. The earlier records span 1983–2023 and lie outside 100 km in the distance check. One USGS-selected boundary event is approximately 200.016 km away under the separate spherical calculation; it was retained and marked rather than silently excluded.
At the lower threshold of M ≥ 4.0, the 100 km query contains the 38 events examined for depth metadata, all within 48.582 km of the centre. The first and last are about 414.889 days apart. Among the M ≥ 4.5 events, the longest gap is 99.853 days. The sequence therefore should not be described as uninterrupted activity across the entire interval.
What can be reproduced now
The offline evidence package contains the selected official responses, event and depth tables, source URLs, timestamps, hashes and a Python standard-library verifier. It reproduces the metadata comparison and frozen catalogue counts. It cannot validate physical depth resolution or waveform modelling. The catalogue magnitudes also include two magnitude types, mb and mww, without conversion to a common scale.
Off-ridge seismicity in the broader region has an existing literature, including Bergman's 1984 study. The original candidate's 2,997 comparison-region definitions and exact referenced Figshare study were not recovered. Together with changing catalogue completeness and the selected query windows, that gap prevents a significance or new-mechanism claim. The supported result is a traceable description of this sequence and how its depth values were recorded.
Sources and assistance
The USGS origin-field documentation, product-selection documentation, first event and 9 km event provide the primary source trail. The source observations are credited to USGS and its contributing observers. The metadata check was completed on 9 October 2026. Byungwoong Yoo directed the comparison; source retrieval, extraction, code and preparation of this account used AI assistance. This is a catalogue case analysis, without journal peer review.