North Korean Nuclear Tests recorded at EPSO

This page records seismic events from the Punggye-ri Nuclear Test Site, and also a natural earthquake that occurred on the Korean Peninsula shortly after the September 2016 nuclear test.

Date Time
(UTC)
Location
(lat, long)
Depth
(km)
Magnitude Range
(km)
Bearing
()
Peak
Velocity
(nm/s)*
Summary
Seismogram
PSN data
file
News
Media
2016-09-12
(Natural earthquake)
11:32:54
35.769N, 129.227E
10
5.4
7700
342.7
73
S6000 triaxial
EPSO_Array
psn
psn
1
2016-09-09
(Explosion)
00:30:01
41.298N, 129.015E
≈1
5.1
8290
344.3
75
S6000 triaxial
EPSO_Array
Willmore Z
psn
psn
psn
1 , 2
2016-01-06
(Explosion)
01:30:01
41.309N, 129.034E
≈1 4.8
8290
344.4
36
S6000 triaxial
EPSO_Array
psn
psn
1
* using the vertical component from the Sprengnether S-6000 seismometer


Seismometer observations:
The first North Korean underground nuclear explosion (UNE) to be clearly observed at EPSO was detonated on the 9th September, 2016 at 00:30:01 UTC.  The earthquake equivalent magnitude for this event was 5.1 (Mb), and 702 seconds following the explosion a weak brief seismic signal was detected at EPSO, being the first few cycles of the P-waves.  Following this unexpected detection, earlier EPSO data were checked and it was found that P-waves had also been recorded from an earlier North Korean test that occurred on the 6th January, 2016.  The table below shows seismograms from the two 2016 nuclear tests as recorded by EPSO's Sprengnether S-6000 triaxial seismometer, which show that for both events the signal amplitude was not much above the background noise, and that the local ground motion primarily vertical in orientation.  The P-phase arrival times (i.e. the 'P'-markers shown on these plots) were estimated by the IASP91 velocity model which is incorporated into the WinQuake seismic analysis program which produced these plots.  No other longer period waves were detectable above the noise floor.  The frequency of these P-waves was about 1.8Hz (6th January) and 1.5Hz (9th September).

6th January, 2016
9th September, 2016
60 seconds 60 seconds


EPSO Array observations:
The P-waves from the 2016 North Korean tests were also observed by EPSO's 3-element seismic array, which are presented in the table below.  This array incorporates three Kinemetrics SS-1 Ranger sensors (in vertical orientation) which all clearly detected the signal, and in near-perfect synchronisation.  This implies that the P-waves are traveling essentially perfectly normal to the Earth's surface, and explains why the local ground motion is predominantly vertical.  This also implies that the events were teleseismic and not due to local noise.  As with the Sprengnether S-6000 sensor, these P-waves had frequencies of about 1.8Hz (6th January) and 1.5Hz (9th September).

6th January, 2016
9th September, 2016
60 seconds 60 seconds


Comparing UNE's with a natural earthquake:
By a very peculiar coincidence, a natural earthquake of magnitude 5.4 occurred in South Korea around three and a half days following the September 2016 nuclear test, which serves as a useful tool to compare UNE's and natural earthquakes.  The table below presents comparison 60-second seismograms from the September 2016 test and the following natural earthquake, from both the S-6000 seismometer and EPSO Array sensors..


Triaxial seismometer (S-6000)
EPSO Array (SS-1 Ranger - vertical)








M5.1
Nuclear
Test









M5.4
Natural
Earthquake


60 seconds 60 seconds

The table above shows some noteworthy differences between these two events.  The triaxial seismometer traces (left hand column) suggest that the nuclear test waves cause primarily a vertical motion of the ground at EPSO, whereas the natural earthquake waves appear to generate more horizontal ground motion.  Also the waves from the nuclear test immediately reach maximum amplitude and then tail off, whereas the natural earthquake waves increase and decrease in a more haphazard manner.  For both the nuclear test and the natural earthquake the peak spectral amplitude of these initial P-waves was around 1.5-1.8Hz.


Initial compression of UNE P-waves:
Another indicator that the nuclear test 'earthquake' was indeed caused by an underground explosion is to look very closely at the very first seismic waves to arrive.  At the point of an underground explosion (both nuclear and mining) the earth is compressed outwards away from the explosive charge, causing the initial seismic wave to convey a motion away from the blast.  So for upward traveling seismic waves originating from a distant blast location, one would expect the initial compressional wave to cause the ground to move upwards.  The adjacent plot shows unfiltered data from an EPSO Array seismometer (September 9 event) and shows that the very first indication of the P-wave's arrival was indeed in an upward direction (i.e. upwards on this plot and also for the actual ground).  Seismologists call this effect compression and the opposite motion is known as dilation, and for natural earthquakes the initial P-waves may show either compression or dilation.




Observations by other Australian seismic stations:
The September 2016 nuclear test P-waves were also detected across various Australian seismic networks, with the adjacent plot showing the propagation of the waves across seismic stations located in South Australia (image courtesy of Geological Survey of SA and Geoscience Australia networks - click on image to view full-size).  The waveforms of the adjacent image look similar to those recorded at EPSO and may also be compared with other example underground nuclear test waveforms presented in Datasheet DS 11.4 of the New Manual of Seismological Observatory Practice.


So are the North Koreans really detonating UNE's?
For smaller seismic events, say less than magnitude 4, it is feasible to simulate a UNE by detonating a sizable quantity of mining-grade explosive down a deep mine shaft, and it is difficult to unequivocally prove what caused the blast simply by looking at seismic signals alone.  Chemical explosions in mines may in a few extreme cases generate earthquakes up to ML = 4.5, but for man-made seismic events larger than magnitude 4.5 the cause is almost certainly from UNE's.

Making your own observations:
The purpose of this web page is to catalogue EPSO's observations of North Korean nuclear tests, but also to demonstrate to interested amateurs that it is perfectly possible to detect relatively small UNE 'earthquakes' even at large distances (8300km in this case) using modest economical seismic logging equipment and home-built or surplus seismic sensors.  Having said that though, the two essential requirements are the use of quality sensors, and operating at a seismically quiet location.  When operating within or nearby towns and cities, the very weak UNE seismic signals are very likely to be swamped by vibrations caused by automobile traffic, trains, etc.

The raw unfiltered PSN data show that the weak North Korean UNE signals may at first glance appear to be completely buried within microseismic and local noise, but since the waves arrive at EPSO with a relatively uniform frequency of somewhere between 1.5 to 1.8Hz, applying a 0.5-4Hz band-pass filter to the raw data very effectively pulls the signals from the noise.  The great circle distance from the Punggye-ri Nuclear Test Site to London, England is around 8600km, and likewise the distance to San Francisco, CA is around 8750km, and it is 10,750km to Washington DC.  So Europe and North America are similarly distant to North Korea as EPSO is, and therefore interested amateurs should be able to detect Korean UNE waves with comparable amplitudes to EPSO (although there is a global P-wave shadow zone at great circle ranges of around 11,400-15,800km, from where P-waves cannot be received).

Wikipedia's list of North Korean nuclear tests presents the explosive yield of the January 2016 test at 7-10 kilotons and the September 2016 test at 20-30 kilotons, so assuming these estimates are correct then the September blast was around thee times more energetic than the January blast.  The table at the top of this webpage logs the peak velocity of the S-6000 seismometer, at 36nm/s for the January 2016 test and 75nm/s for the September 2016 test, so the velocity amplitude increased by a factor of around 2.1 between these two tests.  It is probable that the relationship between explosive yield and EPSO's peak amplitude will follow some kind of power-law of the kind Yield=A(peak velocity)B.   Supposing this is true and then solving for A and B, and we find that for a maximum explosive yield measured in kilotons and EPSO ground velocity measured in nm/s, A=0.0468 and B=1.50, so our equation may be written...

Yield = 0.0468 (EPSO peak velocity)1.5

The yield values used to derive A and B were Wikipedia's maximum yield estimates, so the lowest yield estimates will be around 70% of the maximum figures.  The previous equation may also be rewritten to give the estimated EPSO ground velocity as a function of yield, as...

EPSO peak velocity = (yield/0.0468)-1.5

Using this expression we may roughly estimate the peak ground velocity at EPSO for the three earlier tests that were not recorded.

Test number
Date
Time
(UTC)
Yield kilotons
(ref Wikipedia)
Estimated EPSO peak
ground velocity (nm/s)
Measured EPSO peak
ground velocity (nm/s)
Earthquake
magnitude
1
2006-10-09 01:35:27
0.7 2
12
-

2
2009-05-25
00:54:43
5.4
24
-

3
2013-02-12
02:57:51
16
49
-

4
2016-01-06
01:30:01
10
36
36
4.8 (Mb)
5
2016-09-09
00:30:01
20  30
75
75
5.1 (Mb)

Assuming the estimates shown in column 5 above are correct, EPSO could have probably also detected nuclear test numbers 2 and 3 with the currently installed sensors.  It is interesting to note that the UNE yields have been steadily increasing in size with the conspicuous exception of test number 4 (which was also claimed to be an H-bomb) whose relatively low yield suggests a probable fizzle event.