Nuclear Test Map For 1969 to 2017

By T. McDonald | 29/07/19 | Updated 07/10/19

Did you know that nuclear tests can be felt on the Richter scale? This investigation maps the nuclear tests registering 5.5 or greater on the Richter scale from 1969 to 2017. (Click on the markers to find out more about that test). At first it may look as if there are not many tests, but if you zoom in on one of the clusters, you will see many more markers. During the investigation it was discovered that the cause 'explosion' was actually four more nuclear explosions. Twelve test sites were discovered with six consisting of many tests, and six with less than ten occurrences. North Korea with just one test was the last nuclear explosion detected in the dataset. The average magnitude showed no real difference with nuclear a explosion was 5.86 and earthquakes at 5.87.  Furthermore, the biggest nuclear explosion was in 1973 and took place at Novaya Zemlya, Russia.  Further investigation revealed little difference between the earthquakes and the nuclear explosions except for the depth location.

Aims And Objectives

This investigation uses just one dataset, which covers earthquakes from all over the world from 1st of January1969 - 31st of December 2018 with magnitude 5.5 or greater. The dataset lists causes of the earthquakes in one of ten columns.  The three causes are:
  • Earthquake (22942)
  • Explosion (4 rows)
  • Nuclear explosion (173 rows)
This investigation considers three questions:
  • How many test sites have there been since 1969 - 2019 and in what locations?
  • Which decade had the greatest number of nuclear explosions and what where the biggest?
  • Is there any difference between the nuclear tests and earthquakes in any of the measurements?
No specialist knowledge is required for this investigation and is of use to anyone with an interest in this area.


Ever since nuclear weapons have existed, people have been interested in where the test sites are and who is doing the testing. This investigation examines the data of earthquakes and other cause and puts the locations on a map, discovers the decade with the most nuclear explosions and also examines the difference between the nuclear explosions and earthquakes data.

This report uses just the one dataset that describes all the earthquake data since 1969 to December 2018 with magnitude of 5.5 or greater. One limitation of the dataset is that it does not give any indication of radiation or if there where any effects on the local wildlife. As a result the data will not show if the testing has effected the local wildlife. Furthermore, the dataset does not indicate any air detonations for nuclear explosions; thus, is incomplete.  Another limitation is that the data does not state if a tremor is due to an aftershock or not. Although the dataset is under an MIT license and is copyrighted by Socratica, the license allows me to modify and distribute for private or commercial.

Data Sources

Full details of data sources and licences are given in Appendix 2.

Earthquake Data

The earthquake dataset was obtained from (Socratica 2019) and claims to have data from all earthquakes from 1969 to December 2018 with magnitude of 5.5 or greater. The data is freely and publicly available, but has the condition of the following copyright notice Copyright (c) 2019 Socratica under the MIT licence. Although it was not necessary, it was imported into a SQL database for storage and retrieval for educational purposes.

The data is presented as a single CSV file. Each row in the CSV file represents a single earthquake. The data needed next to no cleaning with no missing data; however, the categories for nuclear explosion and explosion seemed unusual and turned out to be the same. For this reason, the four rows with explosion as the cause where changed to nuclear explosion. No other cleaning was necessary. Furthermore, I have been unable to check the validity of the data with any other similar independent sources; however, USGS is a reliable source.

The earthquake dataset contains 23 119 earthquakes and aftershocks. The data contains no information on personnel involved in the explosions, but does give coordinates, place, date and and time of explosion. 177 of the 23 119 where due to nuclear explosions. The first nuclear explosion took place on the 16th of September 1969 in Nevada USA while last was on the 3rd of September 2017 in North Korea. Nuclear explosions form the main interest of this report. A summary of the nuclear explosions by year is shown in figure 1.

Sumary of nuclear explosions from 1969 - 2017 showing most of the occurences happening in the 70s and 80s
The use of the entire dataset was needed to produce a deeper report; however, subsections of the report where needed to answer some of the questions. For instance, the 177 explosions were a subset used to answer the questions pertaining to which decade had the most nuclear explosions in the study (see table 1). In contrast, the question pertaining to examining the difference between the measurements for nuclear explosions and earthquakes required to whole dataset.

Number of Nuclear Explosions by Decade 

Table 1 showing 1980s with 93 nuclear explosions, 1970s with 70, 1990s with 12, 1969 with 1, 2010s with 0 and 2000s with 0
Table 1 shows the 80s with the most occurrences

A subset was used to examine the number of explosions for each area. In addition, I noticed that the locations in the USA where all in a very similar area; for this reason, I created a single entry 'National Security Site, NV' for any site with Beatty, NV in the place column to produce table 3 below. This consisted of consolidating 8 rows and giving it the label National Security Site, NV. See table 2 and 3.

Before merging

Table 2, before merging of Beatty, NV rows.

After Merging  

Table 3, after merging the Beatty, NV rows in table 2.
Furthermore, when producing table 6 'Differences between Earthquakes and Nuclear Explosions' the values where rounded to 2 d.p.

Analysis Pipeline

Earthquake Data import and preparation

The earthquake dataset was imported into an in-memory data structure for initial analysis with pandas. The data did not require any particular cleaning before use; however, i did make many different DataFrames and reshape them in order to answer the questions. Since this report is also about learning SQL, the data was imported and stored in a pgAdmin local database on a local server for investigations using postgres as the SQL engine and examined with SQL alongside pandas. This involved creating a database with appropriate columns then importing the earthquake csv file.

After initial investigation, it was obvious that the data required shaped to fit each question in this report and aggregate data extracted where necessary and results were used to create charts to illustrate the findings. To illustrate the location findings of the nuclear explosions, it seemed best to use a map. The map was created and markers that give more information added using a simple algorithm after the data for explosions and nuclear explosions was extracted and reshaped to form its own table.

There was no need for analytical algorithms.


How many test sites have there been since 1969 - 2019 and in what locations?

The investigation uncovered 3 sites with more than 20 occurrences, 3 sites with between 5 and 20 occurrences and 6 sites with only 1 occurrence.

The USA had one site area with 8 tests in Nevada covering area 18 and 19 while the ex-Soviet Union had many locations with numerous occurrences. Notably, the highest number of tests was in Eastern Kazakhstan with 93 occurrences. The next highest number of occurrences at a site was in the French Polynesia region with 30 occurrences. This was followed by Novaya Zemlya, Russia with 24 occurrences, which was spread over two sites. Next came Southern Xinjiang, China with 10 occurrences, National Security Site, Nevada with 9 occurrences and Russia-Kazakhstan border region with 6. The remaining 5 location had only one entry each. See figure 2.

Bar chart with Eastern Kazakhstan at 52.54%
Figure 2 shows Eastern Kazakhstan with 52.54% of the occurrences.

From figure 2 it is possible to see that Eastern Kazakhstan accounts for 52.54% of all the entries for nuclear explosion in the dataset. Occurrences for nuclear explosions for Eastern Kazakhstan ranged between 1973 and 1989 and had an average magnitude of 5.40. The top years for occurrences where 1984 with eleven and 1987 with ten. See figure 3.

Figure 3 shows most nuclear explosions for Eastern Kazakhstan took place in 1984

Although figure 2 gives good visuals on the number of nuclear explosions, a map gives a much better idea of locations, see map 1 below.

You may not see all the clusters of markers in map 1, so move the map around to find the different clusters in different locations then zoom in on any of the clusters and select a marker for more detail such as magnitude.

Map for nuclear explosions from 1969 - 2019

Map 1, the original map is hosted at sciencetony.github.

Which decade had the greatest number of nuclear explosions and what where the biggest?

Most of the nuclear explosions in this investigation happened in the 1970s and 80s. The 1970s accounted for 39.55% of the total entries for nuclear explosions with this percentage increasing to 52.54% in the 1980s. See figure 4.

Figure 4, shows most occurrences centred around the 1970s and 80s.

It is worth noting that the total percentage for occurrences in the 70s and 80s adds up to 92.9% of all occurrences. It is also worth noting that in recent years North Korea is the only nation to make a test, which weighed in at 6.3. In fact, the last occurrence before North Korea was in 1996 in southern Xinjiang, China. The occurrence with the highest magnitude for the period between 1970 and 1989 was 6.9 in Novaya Zemlya, Russia on the 27th of October 1973 and was the biggest for all nuclear explosions in the dataset. In addition, Novaya Zemlya, Russia had the top four highest magnitudes for nuclear explosions for the whole dataset and shared fifth place with southern Xinjiang, China (1992). In fact, Novaya Zemlya, Russia had six out of ten highest magnitudes for nuclear explosions for the full date range. Although Eastern Kazakhstan had the most occurrences and had four of the top ten occurrences, Novaya Zemlya, Russia was the place for the biggest explosions. See tables 4 and 5.

Top 10 Magnitudes for 01/01/1970 - 31/12/1989 with cause of Nuclear explosion

Table 4, shows the top 10 highest magnitudes for the date range

Top 10 Magnitudes for 1969 - 2019 with cause of Nuclear explosion

Table 5, Novaya Zemlya, Russia has the top 4 positions and all in the 70s

Is there any difference between the nuclear tests and earthquakes in any of the measurements?

I looked at whether magnitude could shed some light on the matter, but i quickly realised that i was going to need more. Even aggregate data such as maximum, minimum and average yielded nothing; however, i found something of use once i looked at the depth. As can be seen from figure 5, there is a massive difference between depth. While the average depth for earthquakes is 71 km, it is almost non existent for nuclear explosions.

Figure 5 show the advantage depth as the only real difference between earthquakes and nuclear explosions. 

 Differences between Earthquakes and Nuclear Explosions 

Table 6, aggregate data for earthquakes and nuclear explosions

Although the dataset gives no indication for the measurement of depth, the USGS (1989) States: 'Earthquakes can occur anywhere between the Earth's surface and about 700 kilometres below the surface.' The depth range in this dataset is between 0 - 700; for this reason, it is assumed that the depth is measured in kilometres.  There was also a difference in locations with only a few nuclear explosions anywhere near earthquake locations. China and North Korea, however, did have nuclear explosions in a similar region to places that have had earthquakes.  Subsequently, it is possible to use location as an indication of cause.  See map 2.

A Map Showing The Different In Locations For Nuclear Explosions And Earthquakes

Map 2, green markers = earthquakes, purple markers = nuclear explosions. The original map is hosted at sciencetony.github.


Map 1 shows 12 clusters of occurrences with ten of them taking place in the eastern part of the world, which includes the area with the largest number of occurrences: Eastern Kazakhstan with 93 shown in figure 2 and table 3. Eastern Kazakhstan had its larges number of occurrences in 1984 and 1987, see figure 3.  However, Novaya Zemlya, Russia had the biggest magnitudes, which was also the location of the nuclear explosion with the highest magnitude, see table 5, which was 1.03 above the average for the dataset, see table 6.   Notably, the largest volume of data occurred in the 70s and 80s, which accounted for almost all of the occurrences in the dataset with 92.9% shown in figure 4.  As for the difference between earthquakes and nuclear explosions, while earthquakes occurred on average at a depth of 71 km, nuclear explosion occurred at around 0.28 km depth on average.  It was also noted that the location of the occurrence give some indication of the type of cause, see map 2.

To reiterate, there were 12 sites with the greatest number of nuclear explosion occurring in the 70s and 80s with the biggest taking place in Novaya Zemlya, Russia.  Ten sites were in the Eastern part of the world with the USA and French Polynesia region accounting for the other two.  In adition, the biggest took place in Novaya Zemlya, Russia in 1973, but most occurrences took place in Eastern Kazakhstan over all with 93. Depth and location were the only differences between earthquakes and nuclear explosions.


USGS (1989) 'Determining the Depth of an Earthquake [Online]. Available at (Accessed on 20/07/19).

Appendix 1: notebooks

Full details of the data processing and analysis are given in the following Notebooks.
  1. Examine earthquake with pandas. - Investigation into the number of occurrences over time and location.
  2. SQL_Pandas_and_sqldf. - SQL With Pandas And Making Maps With Folium. Investigation into location of occurrences and the making of the map. Note this notebook also includes the making of an earthquake make, which i did not use for this report as it yielded no information for the questions.
  3. Examine earthquake using SQL magic. - Using SQL (postresql and magic) to examine earthquake dataset. This includes aggregation and examples of how to use SQL postgres.
  4. Create A View And Find The Difference Between Nukes and Quakes. - Creating A VIEW and using pandas to plot charts. This notebook covers the investigation into differences between nuclear explosion and earthquakes. It also covers aggravate data.
  5. SQL with Pandas Data Range EK and NK. - Covers the EK plot and analysis and N Korea
  6. Investigation of 70s and 80s and biggest nukes - Creating dataframe for all occurrences between 1970 - 1980. Look at the complete range and find the biggest magnitude and its location

Appendix 2: data catalogue 

The earthquake data in this report came from the USGS (United States Geological Surveys) Earthquake Hazards Program hosted on the

The dataset was provided by Socratica and hosted at under an MIT licence with a copy right notice (Copyright (c) 2019 Socratica), which allows modification and distribute for private or commercial use.

The earthquake dataset is available at [Accessed on 12/7/19]

Appendix 3: evidence for counting the 4 explosions as nuclear explosions

Appendix 4: additional areas to explore

After looking at the data i feel another investigation into earthquakes and depth would yield some interesting information. For instance, while investigating the locations of nuclear explosions i created a map that showed the locations of earthquakes too. This showed that the earthquakes appeared to occur in lines, which could reveal the locations of the fault lines.


Popular posts from this blog

Random Images from API with python

How to get started with Python programming

Build A Test Website In 3 Easy Steps