2. INTRODUCTION
This document is a report on the analysis of the numerous noticeable/ unnoticeable Tremors
that have occurred across the globe during the month of November, 2015. The earth, albeit
seeming placid and immobile, is a very dynamic being. The continental plates continue to move
with vast momentum that causes a rubbing/friction in the plates, resulting in Earthquakes.
Although tremors almost always occur every day, most of them go unnoticed since it causes no
damage to life or property whatsoever. A tremor although almost always being caused
naturally, sometimes are caused by human activities like Mining, Chemical Explosion, etc.
Deadliest earthquake in recent history
2004 – Magnitude 9.1 - A massive earthquake just off the West Coast of Sumatra in the Indian
Ocean. The epicenter of the tremor being underneath the ocean, huge tsunami waves were
generated that traversed the entire ocean, affecting not only Indonesia but also India and Sri
Lanka. The U.S. Geological Survey estimated that a total of 283,100 people had lost their lives,
14,100 went missing and 1,126,900 were displaced.
2015 – Magnitude 7.8 – Nicknamed ‘Terror on Everest’, it is one of the worst quakes to have hit
humankind. This earthquake brought an entire nation to its heels and changed the topography
of the world forever and Carto graphs had to be redrawn.
2011 – Magnitude 9.0 - Near the East Coast of Honshu, Japan. This Earthquake is most known
for the Fukushima Nuclear Leak, which is said to be still affecting lives across Japan.
2007 – Magnitude 8.0 - Near the Coast of Central Peru
2008 – Magnitude 7.9 - Eastern Sichuan, China
Information about the Dataset
The dataset consists of details of all measured tremors around the world for the month of
November, 2015.
(Source: U.S. Geological Survey, http://earthquake.usgs.gov/earthquakes/feed/v1.0/csv.php)
The dataset consists of 7874 records with 17 columns providing details about the tremors.
Columns
Time: Time and Date of Occurrence Gap: Gap between adjacent stations
Latitude Dmin: Dist. between station & epicenter
Longitude Rms: Root mean Square of prediction
Depth: Depth below the surface Net: Network source of information
Mag: Magnitude in Richter Scale ID: Tremor ID
MagType: Type of Measurement Updated: Information Updated on
Nst: Number of Seismic Stations Place
Type: Cause of the Tremor
3. Objective of Study
To find correlations among the variables involved in measurement of a Tremor and to provide
visualizations to understand the data better.
ANALYSIS AND VISUALIZATION
The dataset contains non-normalized data which can be normalized as follows:
The following columns can be recorded in separate tables and referred to the main table
using their respective identities as Foreign keys in the main table
Cause of tremor – Divided into various classes based on records.
MagType
Net
The country of occurrence can be extracted from the field ‘Place’ and recorded into a
separate table which can create two more Normalized tables (as shown in Figure.1)
FIGURE.1
DATA CLEANING
Since the dataset is well organized, not much cleaning is required. Modifications done:
Some column names were changed so that they would be easy to work on
Mag – Magnitude
ID – TremorID
Updated – UpdatedOn
The Country of origin is extracted from the field ‘Place’ using the delimiter ‘,’
An extra column ‘USState’ was added to the dataset so that state-wise analysis of the
data can be performed for the data
4. PRELIMINARY ANALYSIS
(The data was segregated using MYSQL and analyses were performed in R. All visualizations
have been created using Tableau. Please refer to the appendix for the code.)
1. Places with the maximum number of tremors by magnitude
Most of the earthquakes go unnoticed when the magnitude is lesser than 5. Tremors are felt
above 5 on the Richter scale and necessary precautions are taken to avoid losses.
Figure.2 provides information on the top 10 countries that experienced tremors of the highest
magnitudes in the month of November, 2015. A color-wise comparison provides the number of
recorded tremors for these particular set of countries.
FIGURE.2 –Tremors by Magnitude (Descending)
5. 2. Map with location of tremors
FIGURE.3 Global plot of Tremors
Upon geocoding and plotting the locations of the tremors on a global map, we see that most of
the tremors have been recorded on the boundaries between the continental plates.
The most active regions seems to be the boundary between
1. The Eurasian plate and the Pacific Plate
2. The North American plate and the Pacific Plate
These two boundaries alone contribute to 91.18% of all Tremors in the world.
The Fault line between the Indo-Australian plate and the Eurasian plate, albeit suffering from
fewer number of quakes, experiences tremors of the highest magnitudes. This is a primary
reason for the higher degree of destruction recorded when a quake of huge magnitude strikes
the Indian and Australian region.
(Refer Figure.6 appendix for a map showing the continental plates and their fault lines)
6. 3. Testing for Relationship between Depth of epicenter and Magnitude of Tremors
Earthquakes usually originate well underground. Seismographs are used to measure the
magnitude of the tremor and they are placed at various stations across the world to arrive at a
precise value of the magnitude. Since these seismographs depend on the vibrations along the
surface of the earth and the origin is several miles below the surface, most vibrations are
mitigated.
A hypothesis test is performed to analyze the relationship between the two variables.
Variable 1 – Depth, Variable 2 – Magnitude
Null hypothesis: There is no linear relationship between the two variables
Alternate hypothesis: There is a statistically significant linear relationship between the two
variables.
Test Statistic and Decision Rule:
The F statistic of the observed values is considered as the test statistic. If the p-value of this
observed F-statistic is lesser than the 5% significance level (or) if the observed F-value is lesser
than the critical value of F at 95% confidence (5% significance, 1 and 7873 degrees of freedom),
then the null hypothesis will be rejected. Else, we will fail to reject the null hypothesis.
Calculation and Conclusion:
We obtain a p-value <0.0001 which is much lesser than the 5% significance level. Hence we
reject the null hypothesis and conclude that there is a statistically significant linear relationship
between Depth and Magnitude.
However, the relationship is very weak with an R-square value of 0.3172. Upon analyzing more
data (eg. 10 years of tremor data with Depth and Magnitude), it will be possible to conclude
decisively if the relationship is stronger.
FIGURE.4 Regression Plot – Depth vs. Magnitude
7. 4. Analysis specific to the United States
The North American continental plate, although not as violent as the other tectonic plates,
experiences a lot of minor tremors every day. This is attributed to the active movement
between it and the Pacific plate. Hence, it requires a further detailed study to understand which
states are prone to a large part of the tremors.
FIGURE.5 Heat map of US States by Order of Tremors
As seen from the heat map, most of the quakes are centered along the west coast of the USA
(with Alaska bearing the maximum brunt) which again fortifies the fact that the North American
plate and the Pacific plates are rubbing violently against each other.
5. Other Information from the data
Though most of the tremors are produced by natural causes, some tremors are man-
made. These records can be considered as outliers. In the dataset, barring a few records,
all are naturally caused quakes/tremors.
Cause Of Tremor Percentage
Natural 98.43
Man-made Explosions 0.015
There are various types of measurement of the magnitude of a tremor
Type of Measurement Range of Magnitude Number of Measurements
Md -0.77 to 4.2 2180
mwc 5.1 to 5.1 1
8. Type of Measurement Range of Magnitude Number of Measurements
mwr 3.8 to 5.5 25
Mb 0.9 to 5.9 594
Mh 0.4 to 0.6 2
Mw 3.1 to 4.33 4
mb_lg 2 to 4 119
Mww 5.2 to 7.5 19
Mwb 5.6 to 6.5 6
Mwp 5.8 to 6.3 4
Ml -0.8 to 4.6 4920
As we see from the data, ml – ‘Local Magnitude’ – It is the primary method of
measurement used by Richter and Gutenburg. It hold good for magnitudes lesser
than 5
The next most popular method is md – ‘Duration Magnitude’ – The magnitude in this
case is calculated using the time for decay from vibrations of a seismogram. This
method also holds good only for non-violent earthquakes
For higher magnitudes, the data suggests that novel methods such as mb/mb_lg -
‘Body Magnitude’, mwp/mwb/mww/mwc/mwr – ‘Moment based magnitude’ are
used to precisely get the value of magnitude
CONCLUSIONS
Most of the tremors occur on the boundary between the North-American and Pacific
plates
The Indo-Australian fault line experiences tremors of the greatest magnitudes
There exists a positive linear relationship between the Depth of epicenter below the
surface and the Magnitude of the tremor
In the United States, Alaska is most prone to tremors, followed by California and Oregon
Though most of the tremors are caused by natural movement of the earth’s tectonic
plates, some of them are caused by human intervention
9. APPENDIX
CODES SQL/R
tremor <- read.csv("D:/UC subjects/Data Management/Final
Project/Tremor.csv",stringsAsFactors = FALSE,header=TRUE)
library(sqldf)
library(RODBC)
tremorsqlcorr <- sqldf("SELECT
Latitude,Longitude,Depth,Magnitude,Nst,Gap,Dmin,Rms FROM tremor")
pairs(tremorsql[,-1], gap = 0, pch = ".")
#magnitude is greater than 5 - countries
bigmag <- sqldf("SELECT Country
FROM tremor
WHERE Magnitude>=5
")
#Test for linear relationship
depmag <- sqldf("SELECT Depth, Magnitude FROM tremor")
testing.lm = lm(Depth~Magnitude, data=depmag)
summary(testing.lm)
reg1 <- lm(Depth~Magnitude, data=depmag)
par(cex=.8)
plot(Depth~Magnitude, data=depmag)
abline(reg1)
SELECT MagType,MIN(Magnitude) as MinMag, MAX(Magnitude) as MaxMag
FROM dbo.Tremor
GROUP BY MagType
FIGURE.6 Tectonic Plates and Fault Lines