Significant Earthquake Visualization


Visualization

Ratirat Osiri (Rey)

 

SU-LIS-658-01: Final Project Report

 

In this final project, I chose to work with earthquake dataset. Earthquake is one of the deadliest natural disaster and is really unpredictable. However, I believe earthquake data is one of the most detailed recorded data and is at least one thing that everyone share the common awareness. I also believe that this visualization will be educational and informative for those who interested, and it would give the clearer picture of the earthquake’s trends and how it affects the lives.

 

Process

 

There are actually many earthquake dataset available on the internet, especially from the earthquake or geology organization websites. It took me quite some time to seek the suitable one. I finally picked the significant earthquake dataset from NCEI (National Center for Environment Information) section in NOAA (National Oceanic and Atmospheric Administration) website . I tried to find other dataset to enhance my work along with working with this dataset, but I couldn’t find it. I worked on this dataset with Tableau, although the earthquake dataset is more like location-oriented data, but there are also other details that I would like to displays. Tableau is also much easier to work with than CartoDB in my opinion. However, as I worked on the visualization I realized that not all the details are included in the dataset. I don’t think that this dataset is perfect to in term of completion. A lot of columns and rows have null number, some columns barely have any value number, many data in each row are also missing. I was trying to search for more earthquake dataset to increase more details I could add into workbook, but couldn’t find any proper one so I decided to work with just one single dataset.

 

The first worksheet I decided to work on is the map with magnitude intensity. I use the same settings with my previous work with Tableau, Longitude as column, and Latitude as row. I picked an average ‘intensity’ values as a main value to display on the map and I put both color and size options to display the intensity. The more average intensity, the bigger and darker the circle will become. I struggled with the color and design choice a little because of the overlapping on some areas. The areas with more records tend to get really clustered and become just a thick layer of red color. I ended up reducing color transparency to 50% and used to red outline so the users could see the trends in some areas. I think it is interesting to see the cluster of circles that appear on the map are aligning in the identical line with the earth plate joints, especially on the ring of fire, as seen in figure 1 and 2.     

Figure 1   

Sheet 3

 

Figure 2 (courtesy of NOAA)

 

The second worksheet is the graph comparing deaths and injuries by country. I think the information in this dataset might not be totally accurate since many data are missing. At first, I tried to chronologically display the total deaths and injuries by year, then I realized that it wouldn’t make sense in the way how earthquake trends work, unlike the crime rates, unemployment or medical statistics.I tried to combined two graphs together into stacked graph, but the professor suggested that the structure of the dataset did not allow me to do it. However, I successfully made it with bullet graph style, as shown in figure 3, but I feel like the graph design is not really suited with the data. I eventually changed the graph into two simple bar charts divided into deaths and injuries rate by country by putting country in columns and deaths and injuries into rows. I chose red color for deaths and blue for injuries and sorted the countries by the highest deaths as shown in figure 4. China is the far highest one in both categories, I guess it is because of the colossal number of their population compared to other countries. The dataset also includes total missing people, but I think it would be too much unless I could stack all the deaths, injuries, and missing in one graph. However, dividing graph into two graphs is actually making them more comparable.   

 

Figure 3

Sheet 22

 

Figure 4

Sheet 2

 

 

The third graph is the trend of magnitude since 1900 to present. The dataset itself has a few data recorded since around 2000 BC and I decided to filter all the data before 1900 out. The reason is, the record back into very old period of time is not that detailed compared to the record in modern time. Moreover, I think a lot of the ancient earthquakes are missing due to the lack of technology to keep track of them, so I think that I should just focus on the present time. I chose circle views graph style to display the trends of earthquake’s magnitude on each year, putting year on columns and ‘primary magnitude’ on rows. I needed to adjust the circle’s size a little smaller so they wouldn’t overlap each other and make the graph looks like a mess. The result is the graph clustered with small circles, showing the trends of magnitude happened throughout each year, the significant high magnitude will stand out in the graph. I didn’t really intend to make the users see every single detail, but rather let them see the picture of the trends. The graph is shown in figure 5.

 

Figure 5

Sheet 4

 

Finally, I put all the worksheet together in dashboard and I chose all major graph color in red so it will matched. I had trouble properly fitting all the worksheet into dashboard, I tried to adjust dashboard proportion for quite many times until I use the same proportion with the computer screen’s resolution (1920 x 1080) with landscape settings. The dashboard’s scaling made the graphs more clustered than I thought they would, as shown in figure 6 . I also had trouble with the scaling of each worksheet to made them don’t look to clustered.

 

Figure 6

Dashboard 1

 

UX Research

 

For the UX research, I chose to interviews the internet users on two forums, both are science and geology oriented to suit my project topic. My main questions for them are:

  • Can they comprehend this visualization? Are they easy to comprehend?
  • Do they feel like the the visualizations are missing anything?
  • Their opinion on design and color choice?

Even though I didn’t expect to immediately get all the useful feedback I need, but I learned that asking for proper feedback from the internet is not a reliable method. I decided to contact two of my ex-coworkers who have geology-related degree and diploma, all of us used to work together in the same gemology laboratory. I asked them a favor to give me feedback and they were happy to help. I decided to ask two more persons who don’t have a geology background to tested it. I think it would be appropriate to also include the feedback from general users since I tried to made this visualization understandable for general audiences too. I posted two threads asking for feedback and got only one proper feedback from geology subreddit. The other feedback I got from science forum is literally just ‘they look nice’. Fortunately, one user from geology subreddit is kind enough to gave me a proper feedback. I privately sent the link to Tableau public page of my published work to each participants who I contacted personally, with one live feedback facetime with one participant. I just told them that they are earthquake visualizations and let them comprehended the graph and map on their own.

 

From the overall UX research, I got three participants with geology background and two participants from other backgrounds. I summarized all the results and feedback as follow:

 

  • Two participants had trouble viewing a full version at Tableau public on their device/computer and I had to send them screenshot of the dashboard and the individual worksheet to them instead.
  • Two participants couldn’t comprehend the circle views graph (magnitude by year, figure 5) and I needed to explain to them how to read it, both of them don’t have any geology or earth science background.
  • The participant from reddit commented that it would be better if the detail on each earthquake appears when we hover over it.
  • The two participants without geology background tended to comment on just the appearance and design.
  • All of the participants said that the visualization in overall is clear and easy to comprehend. I’m not sure if they just tried to be nice to me.

 

From the overall feedback, there’re not much of the critical feedback or the error detected that I needed to significantly change the design/data. I realized that the dashboard scaling tended to ruin the proportion of the worksheet and I had to adjust the sizes according to how they would appear on the dashboard instead. At first, I set the dashboard with landscape (as shown in figure 6) setting and one participant commented that it would be much more convenient to scroll up and down than scrolling left and right, and I totally agree with him, then I change to portrait setting later as shown in figure 7 as a final render.

 

Figure 7

Dashboard 11

 

Conclusion

 

One thing I like about visualization is seeing the incomprehensible data become the clear picture before me. This visualization was no different to me, they are really informative and even myself have learned so many things from working with this dataset. I chose three different types of graph to exhibited three different facets of the earthquake data and I think each type of graph served the purpose for each type of data and the messages I wanted to convey pretty well. The graph is still not perfect, but I believe if we have more complete and organized dataset, I believe it could definitely make better graphs and maps.  

 

As for the future, I think this dataset and its visualization would be used as a future reference for further study. I also wish that the dataset would be more complete. Even the various earthquake databases are normally refer to each others, but I still have a feeling that the earthquake database from the different geology organizations are still somewhat separated. I might missed it somewhere, but the centralized global earthquake database would be great.

 

Note: The full version can be found here.