EARTHQUAKE CLOUDS AND
SHORT TERM PREDICTION

Earthquake Database Problems

- 5/15/2001 -

When reading the earthquake data "20001206 17:11:06 39.566N, 54.799E, 6.7Mb A (7Mw, 6.9Me, 7.5Ms) Turkmenistan", one may think its errors as small as "0.001" for the both latitude and longitude, and "0.1" for the magnitude according to the International Error Law. However, it is not true because earthquake data are special. In this essay, I will discuss how special earthquake data are, how their errors affect seismological study, and how we can overcome those error problems.

First, Table1 shows19 sets of data from different countries for the Turkmenistan earthquake. Their average, maximum and minimum of latitude, longitude, and magnitude are exhibited in Table 2, whose errors are not small.

One may think that different countries, of course, produce various data. Table 3 reveals a few normal data (not the largest) from the USGS and its two branches: Pasadena Office in Southern California and Berkeley Office in Northern California. The database does not show the differences between those latitude data, and longitude data, but the magnitude data. The USGS uses "MB", and the both branches use "ML". The table exposes a conflict that sometimes there is no big difference between "Mb" and "ML", but sometimes it can reach 1. Moreover, the data of the both branches are also different, and their magnitude difference can reach 1 ML sometimes (See Table 4).

Table 4 shows that the USGS knows the problem, so its experts change data to make a balance, which is respectable. Moreover, the < Earthquake Database System Documentation > of the USGS announces its instrumental data errors as 0.1, 0.2, 0.5, and 1 degree for Rank A, B, C, and D correspondingly (1 degree equals about 100 km), so one should not misread. In addition, what evidence supports the above error ranks needs more study.

Except error, data loss is another special problem of earthquake databases. The first lost data I found from the World Earthquake Database of the USGS was the 5.6 Turkey earthquake on June 1, 1997 [1]. More examples are in Box < Photos & Images >, whose data sources are mentioned. By contrast, the Chinese Earthquake Database lost more than the USGS, which was one main reason why I did not devote myself to earthquake prediction when I was in China.

Second, both data loss and data error affect earthquake prediction seriously, for example, the lost 5.6 Turkey data had bothered me for three years. Moreover, data errors make databases a rubber rule, by which, a correct prediction could be judged as wrong, too. Moreover, to avoid this problem, I have to extend the windows of my predictions or to increase their probabilities.

The both special problems also affect other seismological study, for example, a paper defined foreshock as "M>=2 occurring within 5 km horizontally and 30 days of the mainshock", according to which, it set up a relationship between mainshocks and their related foreshocks [2]. Since the data errors for the paper were far bigger than the definition, the conclusion of that paper may not be meaningful.

Finally, let's discuss how we can overcome those problems. In Essay < An Suggestion Due To Southern Alaska Geoeruption >, I already pointed out that "the measure does not have a uniform standard, like the 1 Kg Mass Standard in Paris or other international standards, to check the measure error", which is the cause of why those errors exist.

To solve the problem, I suggested, "We need a standard earthquake. Due to no nature earthquake is available, we should create an artificial earthquake, whose latitude, longitude, depth, and magnitude can be set up exactly by an underground chemical bomb." Relying on those artificial earthquakes, we will be able to check the errors of earthquake data exactly.

Except the above authoritative method, some rare geoeruptions and earthquake clouds can be employed to check location errors sometimes, for example, the geoeruption in Image 20001120 9:00 [3], exhibited in our homepage before the Turkmenistan earthquake on December 6, 2000, shows its epicenter at about 39.3 N, and 55.1 E. An example for earthquake clouds is in Image 20010103 5:00 [4]. I had pinpointed its epicenter with the arrow and its magnitude 4.3 ten days before the earthquake exactly. However, this method contains error or is not as authoritative as artificial earthquakes.



Table 1: Various data for the Turkmenistan earthquake on December 6, 2000
No. Time Latitude (N) Longitude (E) Magnitude Rank Producer
1 17:11:43.0 39.0 49.0 6.8 Mb A LED
2 17:11:27.1 39.7 51.0 6.2 M A MAD
3 17:11:23.6 40.4 52.3 7.0 Mb A LDG
4 17:11:19.0 40.4 53.7 6.5 Mm A LOR
5 17:11:18.2 40.4 54.0 6.5 M M+ CSEM
6 17:11:16.0 40.0 53.0 6.4 Mm A LIC
7 17:11:13.0 39.4 54.0 6.0 Md M KAN
8 17:11:07.9 39.7 54.9 7.2 Mw M NEIM
9 17:11:07.9 39.7 54.9 7.4 Ms M NEIR
10 17:11:07.3 39.5 55.0 6.7 Mb A NEIA
11 17:11:06.0 39.5 55.1 6.5 Mm A TBGS
12 17:11:05.8 39.7 54.7 7.4 Ms A GSSA
13 17:11:05.8 39.7 54.7 7.4 Ms M GSSC
14 17:11:05.3 39.8 54.5 6.7 M M+ CSEM
15 17:11:04.7 39.8 55.2 7.0 Ms M SED
16 17:11:04.7 39.8 55.2 5.8 Mb M SED
17 17:10:57.4 39.5 56.2 7.0 Mb A SED
18 17:10:57.4 39.5 56.2 6.4 Ms A SED
19 17:10:31.1 39.3 56.86.5 Mb A


Table 2: The Average, The Maximum & The Minimum of Latitude, Longitude, & Magnitude
Item Latitude (N) Longitude (E) Magnitude
Maximum 40.4 56.8 7.4
Minimum 35,3 49.0 5.8
Average 39.5 54,2 6.7


Table 3: The Data Difference Among The USGS & Its Two Branches
Date Time Latitude (N) Longitude (W) Mb(USGS) ML(PAS) ML(BRK)
1994 01 18 0:43:08 34.377 118.698 5.4 5.2 5.5
1994 01 18 15:23:46 34.379 118.5614.84.8 5.0
1994 01 24 5:50:24 34.359118.629 4.2 5.0 4.4
1994 01 28 20:09:53 34.374 118.495 3.8 4.2 4.3
1994 01 29 11:20:35 34.305 118.579 4.9 5.1 5.4
1994 01 29 12:16:56 34.278 118.611 3.3 4.3 4.4
1994 04 06 19:01:04 34.192 117.095 4.44.8 5.0
1994 04 21 16:37:15 36.3 120.427 4.14.5 4.5


Table 4: Good Work, Pasadena Station Reviewed Data
Date Time Latitude (N) Longitude (W) ML(PAS) before ML(PAS) later ML(BRK)
1990 02 28 23:43:36 34.14 117.7 5.2 5.5 6.2
1994 01 17 12:39:39 34.261 118.534 4.5 4.9
1992 08 17 20:41:52 34.195116.8625.3 5.1
1992 06 29 14:08:37 34.106 116.4024.9 5.5
1992 06 29 14:13:38 34.108 116.404 5.4 5.0


References & Images
  1. "My Prediction No. 23 Is Correct" on August 2, 2000
  2. R.E. Abercromble & J. Mori defined. Occurrence patterns of foreshocks to large earthquakes in the western United States. Nature 381, 303~307 (1996)
  3. Fig. 1 Image 20001120 9:00 The 7.2 Turkmenistan Geoeruption
  4. Fig. 2 Image 20010103 5:00 The 4.3 Southern California Earthquake Cloud


Home | Introduction | Publication & News | Predictions | New Predictions | Essays | Links | Contact

Sign Our Guestbook Guestbook View Our Guestbook

Updated: May15, 2001 | Webmaster