Time distribution of Soviet underground nuclear tests (UNTs) carried out at the Semipalatinsk Test Site (STS)

Human factors in the origin-time distribution of Soviet UNTs, or Which numbers did Soviet generals like?

In this report we describe the distribution of months, days, minutes and seconds for UNTs at the Semipalatinsk Test Site (STS). This information was and will be useful in searches of the records of small magnitude UNTs which are still not detected. It also illustrates "Soviet" style of military activity in the former USSR.

We used the last official Russian publication, "Nuclear Tests in the USSR" [1], as a a main source of information about date, subarea and numbers of all UNTs. This publication does not contain information about origin times and magnitudes of explosions. Such information was found in Bocharov et al. [1989], Ringdal et al. [1992], Lilwall & Farthing [1990], and for smal events in our own reports, Khalturin, Rautian and Richards [1994, 2000].

It is known, that a total 496 UNTs were conducted on USSR territory during Oct 1961 - Oct 1990. Among them, 340 were conducted at STS, 39 at Novaya Zemlya TS, and the last 117 were PNEs conducted at many locations in the USSR. This supplement pertains only to the Semipalatinsk TS.


1. Distribution within the year


Table 1. Distribution of UNTs at STS by month
Month Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
STS 8 20 26 29 23 36 33 18 28 35 33 51

For STS, the monthly distribution of UNTs depends less on the weather, than on technical conditions and the "Soviet" factor. For example, January had the fewest tests: only 8 (2.4%). On the other hand, December was the busiest month, with 51 tests (15.3%).

Authorities at a test site had monthly, semiannual and annual quotas. As in any other Soviet organization, they were trying to fill the quota at the last moment. Therefore in June and especially in December there were unusually many explosions --- even though the weather conditions on the Kazakh steppe in December are not good, with heavy snow, wind and temperature as low as minus 25-30 C.



2. Distribution within the month


Table 2. Distribution of UNTs at STS on different days of the month
Day of months: 1-5 6-10 11-15 16-20 21-25 26-30 Total
Number of UNTs at STS 37 45 46 63 56 93 340

Only 37% of tests were made in the first half of the month. The last 63% were made in the second half, mostly during the last five days.

Among 340 UNTs in STS during 1961-1989 only two (0.6%) were made on first day of a month. The largest number of explosions - 31 (9.3%), were conducted on the 29-th day. During the first three days of the month there were only 16 UNTs (4.8%), but during the last three days (28th through 30th) there were 66 tests (20%).

This is a good illustration of the "Soviet" style of operation, which tries to fulfill the monthly quota at the end of each month.


3. Distribution within the day


Table 3. Distribution of UNTs with different hours of the day
Time interval, GMT Subareas
Degelen Balapan + Murzhik
00:51-00:59 1 6
01:00-01:59 8 14
02:00-02:59 17 18
03:00-03:59 58 39
04:00-04:59 66 13
05:00-05:59 20 10
06:00-06:59 22 2
07:00-07:59 5 1
08:00-08:59 4 1
09:00-12:59 4 0
13:00-00:50 0 0

One can see that 57% of all STS tests were conducted between 03:00 - 05:00 am, 78% - between 02:00 and 06:00 am and 93% between 01:00 and 07:00am.

This information proved useful to us in searching seismograms for signals of small previously undetected events.



4. Distribution within the hour


The minute at which a UNT is set off --- the origin time of the seismic source --- does not depend upon any technical or weather factor. The choice of minute was the exclusive power of a Soviet General - the Chief of STS (for all three sub-areas). So the minute distribution depends upon human factors, the favorite numbers of the General who made the final decision.

It is known, that during 1947-1991 ten Soviet general were the the Chiefs of STS. We don't know how many generals occupied this position during 1964-1989 (when almost all UNTs were made) and how long they served. But it is clear that there were four completely different periods of time within which there was a characteristic distribution of origin times within the hour:

To discriminate between these different distributions, we chose the following parameters:

a. percentage of odd and even numbers;
b percentage of total number of events conducted during the first ten minutes (00 - 10 min) and the last ten minutes;
c. how many events have minutes ending on figure 3 or 7 (e.g., 17, 27, 33, 57 min).
It was found the most preferable numbers are 7 and less 3.
4.1. Minutes of origin time distribution for UNTs at STS during Oct 1961 - Dec 1968
During this period at STS there were 62 UNTs with known origin time.

The following minutes were observed more frequently:

58 min - 14 events or 22% (All were made during Nov 1965-Aug 1967).
00 min - 12 events or 18% (All were made during Feb 1962-Dec 1965).
04 min - 7 events .. 11% (All were made during Sep-Dec 1967).
06 min - 4 events .. 6% (All were made during Jun-Sep 1968).
08 min - 3 events .. 5% (All were made during Apr 1967-Jul 1968).
02 min - 3 events .. 5% (All were made during May 1964-Jan 1967).

So, a majority of UNTs (67%) were made in even minutes of the 10-min period, 58 min - 08 min. 39 digits of the 60 available minutes were not observed.
a. Even minutes have 53 events (86%), odd minutes - 9 (14%). even/odd ratio is 6.2.
b. 31 explosions (50%) were made during the first ten minutes; 21 explosions (30%) were made during last ten minutes.
c. Minutes ending on the figure 3 were observed only two times (3%), minutes ending on figure 7 observed only three times (5%)


4.2. Minutes of origin time distribution for UNTs at STS during Jan 1969 - Jul 1972
During this period there were a total of 40 events with known origin time at STS.

The following minutes were observed more frequently:

3 min - 16 events ... 40%
2 min - 4 events ... 10%
33 min - 4 events ... 10%
57 min - 4 events ... 10%
47 min - 3 events ... 7%

47 figures of minutes were not observed.
a. Even minutes have only 9 events (22%) while odd minutes - 31 (78%). Odd/even ratio is 3.4.
b. During the first 10 min there were 23 tests made - 57% During last 10 min there were 6 tests made - 15%.
c. Minutes ending on figure 3 observed at 21 tests - (52%), minutes ending on figure 7 observed at 8 tests - (20%)


4.3. Minutes of origin time distribution for UNTs from STS during Aug 1972 - Aug 1980

A total 103 events with known origin time were conducted at STS during this period.

The following minutes were observed more frequently:

57 min - 26 events ... 25%
27 min - 13 events ... 13%
33 min - 11 events ... 11%
47 min - 9 events ... 9%
3 min - 8 events ... 8%
7 min - 7 events ... 7%
17 min - 7 events ... 7%
37 min - 6 events ... 6%

39 figures of minutes were not observed.
a. Even minutes have only 6 events (6%), odd minutes - 97 (94%)! Odd/even ratio is 16!
b. During first 10 min were made 16 tests - 15% During last 10 min were made 30 tests - 29%
c. Minutes ending on figure 3 observed at 26 tests - (25%), minutes ending on figure 7 observed at 68 tests - (66%) !


4.4. Minutes of origin time distribution for UNTs from STS during Sep 1980 - Oct 1989

A total 94 events with known origin time were conducted at STS during this period.

The following minutes' numbers were observed more frequently:

57 min - 13 events ... 14%
17 min - 7 events ... 7%

Only 20 numbers of minutes were not observed.
a. Even minutes have only 20 events (21%), odd minutes have 74 (79%). Odd/even ratio is 3.9.
b. During the first 10 min there were 11 tests - 12%
During the last 10 min there were 28 tests - 30%
c. Minutes ending on figure 3 observed at 14 tests - 15%,
minutes ending on figure 7 observed at 31 tests - 33% of total number of UNEs during this period.


4.5. Minutes of origin time distribution for UNTs from STS: Summary for four periods of time

Table 4. Some parameters of the distribution of minutes, for four period of time
Discrimination parameters Oct 1961 Jan 1969 Aug 1972 Sep 1980
-Dec 1968 -Jul 1972 -Aug 1980 -Oct 1989
Preferred minute numbers (here are shown only minutes, which were chosen at least 4 times more often than the random probability, 100/60 = 1.66%) 58m-22% 3m-40% 57m-25% 57m-14%
00m-18% 2m-10% 27m-12% 17m- 7%
04m-11% 33m-10% 33m-11%
57m-10% 47m- 9%
3m- 8%
7m- 7%
17m- 7%
47m- 7%
% of tests made at preferred minutes 51 70 86 21
Not observed minutes, % 65 78 65 33
Even numbers, % 86 22 6 21
Odd numbers, % 14 78 94 79
1 - 10 min, % 50 57 15 12
50 - 59 min, % 30 15 29 30
Ending on 3, % 3 52 25 15
Ending on 7, % 5 20 60 33


One can see that:
  1. The first period General:
    a) liked even numbers (86%!);
    b) preferred to "push the button" during the first ten minutes (50%);
    c) didn't like digits 3 and 7 (3-5%% instead of 10% ).
  2. Second period General:
    a) liked odd numbers (78%);
    b) preferred the first ten minutes and didn't like the last ten minutes;
    c) liked digits 3 (52%! instead of 10%) and 7; 40% of UNEs were made at 03 min!
  3. The third period General:
    a) liked only odd numbers (94%!!!);
    b) preferred the last ten minutes (29%) rather the first ten ( 15%);
    c) liked the figure 7 (60% !!) more then figure 3 (25%).
  4. The forth period General worked in a completely different style. Some time in Aug-Sep 1980, a decision was made that origin time can be choose randomly, not only on the minute (00 sec), and the minutes themselves also can be arbitrary. The level of preference between minutes became lower - only figure 57 stood out (14%). Other minutes were more rare - less then 5 %. Nevertheless his preferred numbers were 7 (30% instead of 10%) and to some extent 3 (15%).


5. Distribution within the minute

Before Sep 15, 1980, all tests were made around 00 sec with an accuracy of a few tenths of a second. After Sep 15, 1980, the value of seconds became arbitrary and the military started to vary minutes and seconds, perhaps to make more difficult the detection of explosions. Nevertheless, the distribution of seconds was still far from random. Thus:

Since Sep 15, 1980, there were 94 UNTs at STS with known origin time. The distribution within the minute, rounded to whole number seconds, is as follows:

9 sec - 26 events (28%)
7 sec - 9 events (10%)
8 sec - 8 events ( 8%)
13 sec - 6 events ( 6%)
0, 5, 10, 15 and 17 sec - 5 events each (total 27%)

So within a four second time interval (between 6.5 and 10.4 sec), more than 50% of the tests were carried out. 69% of explosions were made during the first ten seconds, 23% during next ten seconds, and only 8% during the last forty seconds (20-59 sec).

Odd values of seconds were preferred: 69% of odd against 31% of even digits.

6. Origin time distribution of UNTs at the Nevada Test Site

Nothing "interesting" (depending apparently upon the human factors, etc) was found in the origin time distribution of UNTs at Nevada Test Site. We have no comment on the names of these tests.

References

Khalturin, V.I., T.G.Rautian, and P.G. Richards. A study of small explosions and earthquakes during 1961-1989 near Semipalatinsk Test Site, Kazakhstan. Scientific report, 1994.
Khalturin, V.I., T.G.Rautian, and P.G.Richards,. A study of small magnitude events during 1961-1989 on and near the Semipalatinsk Test Site, Kazakhstan. (in press, for 2000).
Lilwall, R.C., and J.Farthing. Joint epicentre determination of Soviet underground nuclear explosions 1973-1989 in Eastern kazakhstan, AWE Report No O 12/90, H.M. Stationery office, 1990.
Marshall, P.D., T.C. Bache, and R.C.Lilwall. Body wave magnitudes and locations of Soviet underground nuclear explosions at Semipalatinsk Test Site. AWRE Report No O 16/84, 1985.
Mikhailov, V.N. (editor) and 14 co-authors. USSR Nuclear Weapons Tests and Peaceful Nuclear Explosions, 1949 through 1990, RFNC-VNIIEF, Sarov, 96 p, 1996.
Mikhailov, V.N.(editor) and 29 co-authors. USSR Nuclear Tests. IZDAT Publishing House, Moscow, 248 p, 1997.
Ringdal, F. Teleseismic event detection using the NORESS array, with special reference to low-yield Semipalatinsk explosions. Bull. Seism. Soc. Am., 80, 2127-2142, 1990.
Ringdal, F., P.D. Marshall, and R.W. Alewine. Seismic yield determination of Soviet underground nuclear explosions at the Shagan River test site, Geophys. J. Int. 109, 65-77, 1992.


[back to the top]
[back to the list of supplementary material]
[to Paul Richards' home page]