Visualizing Spatial and Temporal Trends in Earthquakes
By: Ashley Lu (Project Lead), Presley Moreno, Tianlang(Tian) Ouyang, Huiyuan Li
Introduction:
Every year, 12,000 to 14,000 earthquakes happen in the United States. An earthquake is caused by the skating of the surface of the earth; when two blocks of earth slip past each other at a tectonic plate boundary known as a fault, a buildup of energy is created that eventually releases a seismic wave. Earthquakes come in a variety of sizes: some are so weak that we humans can’t feel them, while others are a powerful force of destruction that results in massive property loss and the loss of human life. For example, the 1906 earthquake in San Francisco was so severe that it burned much of the vital record the city maintained. Property losses were estimated to be 5.15 billion in 2019. Currently, the United States addresses earthquakes by providing education opportunities and updating building codes. If an earthquake strikes, a disaster declaration can be made and FEMA could start by providing funding, personnel, and supplies.
Overview of the Dataset
We obtained our data from USGS (United States Geological Survey), and more specifically, from the USGS Earthquake Hazards Program. This program is responsible for monitoring and assessing earthquakes and their impacts, as well as researching the causes and effects of earthquakes. We obtained three years’ worth of data from 10/18/2018 to 10/18/2021 for earthquakes and seismic activities of 2.5+ magnitude that occurred in the conterminous U.S., which excludes Alaska and Hawaii. Interestingly enough, this data also includes a few earthquakes that occurred in Canada and Mexico.
The data we collected had a lot of variables to explore, and we chose to mainly focus on the following:
- Time: describes when the earthquake occurred in y/m/d h:m:s format
- Latitude: Latitude of the earthquake
- Longitude: Longitude of the earthquake
- Depth: depth at which an earthquake occurs (below the Earth’s surface); the shallower the more damaging
- Mag: magnitude of the earthquake, categorized from 2.5 or less all the way to 8.0 or more; a measure of the size/strength of the earthquake
- Type: type of seismic activity
- HorizontalError: error in the location of the earthquake
- DepthError: error in the depth of the earthquake; error on depth tends to be greater than location
- MagError: error in magnitude of the earthquake
We also used reverse geocoding to get the city and state names from the coordinates, as we were interested in using maps to analyze where specifically earthquakes occurred.
Exploring the Dataset
One of the first things we wanted to know was whether there were any relationships between the variables. We used a correlation matrix to analyze the linear relationships between the numerical variables in our dataset.
We found that most numerical variables had a weak correlation to other variables, with the exception of horizontalError and rms, which had a positive correlation of 0.74.
Another thing we wanted to do was make use of the error variables in the dataset and see how accurate our earthquake data really was. To do so, we used boxplots for horizontalError, depthError, and magError to analyze the error distribution of the earthquakes.
What we noticed was that horizontalError and depthError tended to be small, with depthError having a larger spread. This makes sense since it is harder to predict depth accurately than location. However, what was surprising was the distribution of magError, which had much greater variance compared to the other two measures of error. This could be an indication that earthquake magnitude is harder to measure in comparison to location and depth, and so some of the magnitudes recorded in our dataset might not be completely accurate.
We wanted to also make a distinction between earthquakes and other seismic activities. In our dataset, most of our data describe earthquakes, but we also have entries that record other types of seismic activities.
We can see that besides earthquakes, mining explosions are the second most common type of seismic activity in the conterminous U.S. Other types of seismic activities include different types of explosions and mining-related events. Since these other seismic activities have similar effects to earthquakes despite being caused by different factors, we decided to include them in our analysis of earthquakes. However, we also explored other types of seismic activities, specifically mining explosions and experimental explosions, more in-depth.
Finally, we wanted to get an overall sense of what were the states that had the most earthquakes, and for those states, what was the highest magnitude. We see that as we expected, California had the most earthquakes, and it also happened to have the highest magnitude earthquake. Nevada and Idaho, while having significantly fewer earthquakes than California, both had at least one high magnitude earthquake at 6.5 magnitudes.
Where do Earthquakes Typically Occur?
When we think of earthquakes, we usually think of California as the earthquake hotspot. This makes sense given the numerous faults in California, such as the San Andreas, San Gregorio-Hosgri, and Hayward-Rodgers Creek faults. Through mapping, we wanted to visualize the locations of the earthquakes to see how California compared to other states, as well as if there were other states with significant clusters of earthquake occurrences.
Our analysis shows that not surprisingly, earthquakes occur most often in California. However, earthquakes do not happen most often along the fault lines. Instead, earthquakes happen most frequently at the intersection of Kern and San Bernardino County, near China Lake, California. Earthquakes are also frequent in Tonopah, Nevada. As you navigate around the density map, observe that the color near these regions is shaded more red (higher density) compared to other areas (including the fault line).
Seismic Activities Outside of Fault Lines
During our analysis, we also come across some interesting conclusions with other seismic activities. Mining explosions are clustered around Mountain Iron, Minnesota, and Wright, Wyoming. Wright is surrounded by the mining of coal and uranium as well as oil and gas development, which explains the mining explosions. Mountain Iron has a similar situation to that of Wright.
Minnesota addresses the potential dangers of mining through mine health and safety training in Minnesota State Colleges and Universities. However, this program seems to focus more on overexposure to silica dust, small particles which can cause severe health problems like lung cancer, and powered haulage accidents, caused by equipment such as shuttle cars and other vehicles. Considering the cluster of mining explosions we found in Minnesota, the state should continue to focus on training miners, especially with regard to dealing with mining explosions.
Wyoming has a similar training program in the Northern Wyoming Community College District (NWCCD) Gillette College, providing mandatory training for miners. This program also emphasizes current national trends in mine safety and hazard awareness, which might relate to mining explosions. Wyoming has a history of deadly mining explosions, and so it is imperative that they provide training such that miners are able to recognize and deal with mining explosions.
Interestingly, the dataset shows three experimental explosions off the coast of Florida. It is possible that there is weapon testing in that location, though it is hard to know for sure.
We tend to associate seismic activity with the west of the United States, but we were also interested in looking at which states had the least amount of earthquakes. These locations were probably the least likely to be around fault lines, and we wanted to look into why these earthquakes were caused. We looked into earthquakes in three different locations: Indiana, Massachusetts, and Nebraska.
Indiana had only one earthquake in this time period, which had a magnitude of 3.82. Even though Indiana is in the middle of the United States and is not near any tectonic plates or major faults, there are two main areas of seismic activity; the Wabash Valley Seismic Zone and the New Madrid Seismic Zone. The earthquake occurred in the Wabash zone, which consists of faults deep under many layers of sediment, and its tectonics are not fully understood yet. Despite Indiana not being a place of common seismic activity, once in 1812 there was a large earthquake in the Madrid Zone that reversed the direction of the Mississippi River.
Next, Massachusetts only had one earthquake as well during the time period at a magnitude of 3.6. Similarly, Massachusetts is not near any plates, but fault lines come from the creation of the Appalachian Mountains and many are undetected. Earthquakes are caused by pressure building up from plate movement that is released in these areas.
Finally, in Nebraska, there were three earthquakes and we decided to look at two that occurred in the same area, in the north of the state. The magnitudes were 3.1 and 3.7, and there is actually minimal information about the cause of these earthquakes. These occurred in the Siouxana Arch region, and most earthquakes in Nebraska are caused by natural pressure from faults underground, but overall, there is little known about the geological structure underneath.
There is still much to be learned about how these earthquakes occur. The more earthquakes that occur in the future in these areas of low seismic activity, the better we may be able to understand the geological structure of areas that are not near fault lines.
Which State has Strong Earthquakes?
Contrary to popular belief, earthquakes are typically not deadly in California. In fact, the top three states with the highest median magnitude are actually all located in the east — Florida, Indiana, and Massachusetts, with a median earthquake magnitude of 3.6 or above.
It’s likely that since California experiences such a high volume of earthquakes each year, most of the earthquakes tend to be of low magnitude, lowering the median magnitude. On the other hand, states like Florida, Indiana, and Massachusetts typically experience very few earthquakes, so a high magnitude earthquake would increase the median magnitude significantly. From this visualization, we can see that it’s still very possible that states on the east coast experience strong earthquakes, and so it’s important that people are aware of this possibility. State governments could implement more basic earthquake preparedness measures to address these earthquakes.
When do Earthquakes Typically Occur?
In our analysis of temporal trends in earthquakes, we wanted to determine if there were particular times when earthquakes would occur more frequently. To do so, we analyzed earthquake occurrences by hour and by month.
We found that earthquake occurrences by hour had somewhat of an oscillating pattern, peaking at around 4 AM to 5 AM and 5 PM to 6 PM. However, for the most part, the number of earthquakes stayed consistent for each hour. Seismologists generally say that earthquakes can occur at any time, so it is interesting to see that over the past 3 years, earthquakes occurred mostly in the early morning and evening.
We also wanted to see if an “earthquake season” existed; namely, we investigated if there were months where there were more earthquakes.
Our analysis shows that earthquakes typically occur in July. From 2018 to 2021, 3290 earthquakes occur in July, which largely exceeds the number of earthquakes that occur in other months. In June, 1063 earthquakes occurred. In August, 852 earthquakes occurred.
Our analysis also explores the change of earthquake magnitude over months. We found that the range of the earthquake magnitude is typically between 2.8 and 3. So, the change of the earthquake magnitude is relatively stable. In July, the average magnitude of the earthquakes reaches 3.004, the highest average magnitude throughout the year.
Moreover, our analysis shows how the average magnitude of the earthquake changes over the years. We found that the average magnitude of the earthquake is highest in 2019, which is 2.97875, and it gradually decreases to 2.91085 in 2021.
The scientific research generally disproves that earthquakes occur at a specific time, and suggests that earthquakes can occur at any time. The substantial increase in earthquakes during July prompts further research, and it’s likely that it relates to the two successive large earthquakes that hit near the Walker Lane and Eastern California shear zone regions during July 2019. These earthquakes were of 6.4 and 7.1 magnitudes, respectively, and released several aftershocks along the faults. These earthquakes tended to be smaller and unlikely to be felt, yet were still registered as earthquakes.
Trends in Earthquake Occurrences
Through the above animation, we get a sense of the earthquakes that have occurred across the dates covered in our dataset. It seems that general patterns tend to remain consistent, such as the numerous earthquakes in California, some earthquakes in Oklahoma, and mining explosions in Wyoming.
Overall, it seems as though the number of earthquakes is increasing slightly across the years, with 2018–2019 having earthquake occurrences in the 100s per month and 2020–2021 having earthquake occurrences in the 200s per month, with the exception of some outliers like July 2019 (which we previously determined to be the cause of a spike in July earthquakes) and May 2020. This could be due to better technology in earthquake detection, as previous small earthquakes may not have been picked up. As artificial intelligence continues to grow, future earthquake detection technology might also use machine learning to pick up even more earthquakes.
We also explored trends in earthquake magnitude and depth. It seems that the median magnitude of earthquakes slightly changes over time. The median magnitude falls in the range from 2.5 to 3.0. There are three small peaks of the median magnitude. They occur respectively around September 2019, March 2020, and September 2021. By observing the change of median depth over time, we can find that the peaks of the median magnitude of earthquakes approximately overlap with the nadirs of the median depth of earthquakes. We can therefore form the hypothesis that the larger earthquakes occur at the shallower depths in the earth’s crust.
Conclusion
Earthquakes continue to remain prevalent in our lives. In our analysis, we learn that earthquakes actually don’t necessarily concentrate along the fault lines. Our dataset initially shows that earthquakes happen most often in July, but we discover the spike was due to the shock of a high magnitude earthquake. Earthquakes also tend to have a higher magnitude in the east coast of the United States. Overall, it also seems that the number of earthquakes has increased in comparison to previous years. Aside from earthquakes, we also discovered interesting patterns and trends of other seismic activities. We would like to thank you for your time and hope you have learned something new from this article. No matter where you are in the United States, it’s important to be aware of earthquakes and be prepared in case one occurs.
