Natural Hazard Mapping
Data Exploration of Tsunami & Earthquake
February 10th, 2021|Yu Nakamura
Introduction
As everyone is aware, we all who live on Earth have to deal with natural disasters such as earthquake, volcano activity, drought, flooding, wildfire, landslide, and so on. Among those various disasters, the massive number of lives were taken due to earthquakes. Tsunami which is a Japanese word that translates to “harbor wave” has often caused by earthquakes, and it has been responsible for over 500,000 fatalities across the world.
In this study, I review the damage of earthquakes and tsunami in the world to provide a knowledge of where and how much significant damages were recorded in past years through the interactive visualization with UGSG and NOAA open database.
On top of that, this is from my curiosity, I dive into the details of the one of strongest earthquake “Tohoku earthquake” which occurred on March 11th 2011 in Japan. The earthquake struck below the North Pacific Ocean at an underwater depth of approximately 29km, 130 km east of Sendai, the largest city in the Northern region in Japan, which magnitude 9.0–9.1 was recorded. The Tohoku earthquake caused Tsunami a few hours after main shock, which killed over ten thousand people and more than four hundreds people lost their home. Wikipedia
Data
The earthquake and Tsunami data are available from the published open source of the USGS catalogue and NOAA natural hazard database. The analysis for this study is done by Tableau and Python.
The interactive Tableau dashboard is available from below Tableau Public links.
- World Earthquake Dashboard
- World Tsunami Dashboard
Data Exploration and Analysis
Where did the earthquakes cause? Any trends of magnitude and depth from past records? Which country had the most severe damages?
Some erroneous measurements are mentioned in the catalog before the invention of seismographs in the late 1880s. To eliminate data correction bias for the data exploration in this study, I exclude the events prior to 1990.
A glance at below earthquake density map, we can see that earthquakes locations were well aligned with tectonic boundaries in the world ocean.
Here is the top 15 countries that has high number of earthquake events Magnitude (Mw) > 5.5 were recorded over past 2019 years.
Top 3 countries are all Asian regions: Indonesia, Japan and China where located on near subduction zones of ocean crusts under continental crusts due to plate tectonics.
Above histogram shows that magnitude is roughly normally distributed, and the frequency of range Mw 5.5 to 7.5 are the most common and the events Mw > 8.0 is very rarely happened. Also we see that a lot of earthquakes recorded in our land was occurred at hypocenters at underwater depth of within 50 km. Some earthquake were caused at more than 100km but it was rare to convey seismic waves to the land.
I create the packed bubbles plot to show the comparison of total damage level equivalent to the USD million dollars grouped by countries. You understand the number from Japan is quite large since there were several recordable earthquakes in Japan over years.
Where, What type of factors caused Tsunami? Which country had the most severe damages?
Again, I repeat same exercise as the earthquake to visualize the effect of Tsunami in the world.
As a below total fatalities in each country, Tsunami damages are recorded in the countries along tectonic boundaries, and high fatalities are confirmed in Asian countries such as Japan and Indonesia.
A Tsunami is a series of extremely long waves caused by a large and sudden displacement of the ocean. Here is the breakdown of cause of Tsunami that 76% of Tsunami events were triggered by earthquake, and 4 to 5% of Tsunami were occurred by landslides, volcano activity, and meteorological activities, respectively.
The interesting here from the max height distribution of below plot is that high degree of max wave height > 50m were associated with landslides events, not earthquakes.
If you go to the interactive dashboard (click here) and select the top 2 countries who has highest frequency of Tsunami: USA and JAPAN, you will be surprised by the difference between them. In JAPAN, 94% of Tsunami evets are caused by earthquake which has 5 to 40m wave heights, while USA has a completely different trend that the portion of cause of earthquake goes down to 33%, and other factors such as landslides and meteorological occupied as a larger portions of the Tsunami events, which caused a quite higher intensity and wave height > 50m.
Below packed bubble plot shows the total damage due to Tsunami by countries. Again it is a fair comment to say that Asian countries located along the pacific coast ex. Japan, Indonesia etc. who also recorded severe damage from earthquakes mentioned above, have been listed as the worst severe damaged countries.
Is there any type of earthquake which potentially would trigger Tsunami?
In this section, I would like to dig more about the relationship between earthquake and Tsunami. All agrees that earthquake is the most common trigger for Tsunami. So can we simply say that the more damaging earthquake increase the chance of Tsunami? To understand what type of earthquake can potentially trigger Tsunami, I implemented data analysis with clustering methods.
First, I built scatter plot showing Focal depth vs. Primary magnitude grouped by Tsunami occurrence.
It shows Tsunami tends to occur the earthquake events Mw > 6.0 or 6.5 with shallower focal depth < 50m.
First, I implemented Principal Component Analysis with selected 6 attributes to see whether if any useful grouping can be identified for the prediction of Tsunami occurrence. Damage description is an ordinal variable that corresponds to the degree of damage from 1 to 5 (less =1, severe = 5).
From above plot , primary component vs second component, some trends between the variables are detected:
- EQ_PRIMARY, FOCAL_DEPTH and REGION_CODE are positively correlated
- DAMAGE_DESCRIPTION and YEAR are completely negatively correlated which I think it might be due to data correction bias that more data is available in recent years.
However, unfortunately, I can’t see any good separation which related to Tsunami occurrence “Yes” or “No”.
Next, I tried k-mean clustering with same attributes combinations to see possible clustering which help to understand Tsunami occurrence.
Above plots shows two clustering group which is close to the grouping from the original plot below colored by Tsunami occurrence.
K-mean clustering allows to highlight two clusters corresponding to earthquake that triggered Tsunami or not but it didn’t capture the details. If you see the events with deeper depth that Tsunami weren’t triggered was misclassified as same group as the one which expects Tsunami occurrence. I would conclude that k-mean clustering cannot be an accurate tool to categorize the subgroup whether Tsunami occur or not.
Case Study: What happened during Tohoku Earthquake on March 11th, 2011?
Lastly, I introduce the analysis of Tohoku Earthquake on March 11th, 2011, the fourth most powerful earthquake in the world since modern record-keeping began in 1900.
From below correlation tables for monthly counts of earthquake and Tsunami, as expected, we recognize highest numbers of both events on March 2011 corresponding to the Tohoku Earthquake.
Here is the hourly recorded data of each earthquake features during the Tohoku Earthquake. Before the main shock on March 11th 5:46 am UTC, high magnitude events > 5.0 was increasing suddenly that can be seen as spikes in below plot, which corresponding to the known foreshocks events on March 9th and 10th, respectively. After mainshock on March 11th, we continue to see high activity of earthquake M > 4.5 until March 23rd, few weeks after mainshock. Also the high magnitude peak on April 11th is the one of the events due to aftershocks according to the information from wikipedia.
Closer looking at the events location through time, it is very quiet before March 9th. Increasing the activities from March 9th to 10th that recorded foreshocks events Mw = 7.0, and we see the highest intensity Mw = 9.0 on March 11th morning. The total earthquake between March 11th and 12th is 720 which is 16 times larger than the one in prior two days. After March 12th, we constantly see numerous events with higher Mw events up to 6.0 and the events counts are gradually decreasing to 100 events level per 2 days a week after the mainshock.
Since the lack of time data in Tsunami data record, I only can plot the Tsunami feature during March 11th as daily basis. There is three different Tsunami source was recorded around March and April 2011, which corresponds to the events of foreshock, mainshock, and aftershocks events, respectively. On March 11th, more than 5,000 waves were reported with maximum height is 56m. The average travel time doesn’t vary between each that indicates that those three sources are triggered on east coast of Northern Japan closer to where epicenter of main shock was.
Conclusion & Wrap up
Through this data study, I achieved to showcase the analysis of damage of earthquake and Tsunami in the world and high resolution analysis of Tohoku Earthquake in Japan, March 2011 with usage of Tableau and Python.
There is plenty more to explore natural disaster’s database for example adding the volcano activity data to this study or involve more statistics approaches to earthquake prediction with multiple attributes or improve the accuracy of Tsunami occurrence model etc. I would like to continue learning and developing my skillsets of data-driven analysis and explorations.