Final Assignment – Tornadoes in the US

on 17|02|2015
Filled under: Final Assignment

Final Assignment

Interactive Visualization of Geo-Located Data Over Time

000901

For the final assignment, I have chosen to visualize the tornadoes that have occurred in the United States in the past decades. I have taken interest in this particular area, as I have lived in the mid-west for 10 years, where I have experienced many tornadoes within the vicinity.

The dataset I have used to achieve this can be found here.  This data set is provided by NOAA’s National Weather Service, by their Storm Prediction Center WCM (Warning Coordination Meteorologist). It provides statistics of most tornadoes that have taken place in North America from years 1950 to 2013. Details about each tornado are provided, including – but not limited to – state(s), starting location, ending location, fscale (intensity), time, states affected, property losses, injuries and fatalities, wind gust, length and width, and date.

dataset

The reliability of this data set is high, as it is provided by a governmental weather service. Details describing the data set, including units measured, how to read the information, what the information describes, how it could be used, etc…, are provided with the data set in .pdf form.

pdf

I have visualized this data using a combination of unfolding maps and controlP5 in ‘Processing 2.2.1′.  What I have aimed to achieve with this project was to provide an easily readable visualization of geolocated tornadoes, where the user could efficiently and comprehensively understand and comparatively analyze the tornadoes that have occurred temporally, geographically, and statistically.
First, I have imported the data set in .csv form as a table, where I extracted all relevant information to what I aimed to visualize. I then introduced three interactive controlP5 sliders, where the user has the flexibility to choose the exact day, month, and year they would like to view. The animation loops through each day automatically, but the user has the freedom to slow down the frame rate, fast forward it, and pause it for exploration of tornado incidents on any chosen day. The user also has the freedom to choose amongst three map styles (technical toner, road map, or terrain), zoom, pan, and hover over the any of the plotted tornadoes to view a summary of the corresponding statistics. When a tornado is hovered over, it is highlighted with a bold circle, and the data appears on the left-hand side of the screen, complete with units and headers of what each piece of information stands for.

16

One constraint of the dataset is that some ending locations of tornadoes are unavailable due to lack of recorded information. To maintain accuracy of the visualization, I have symbolized this using a central yellow-colored highlight for each of these tornadoes, using conditional statements.
20

Having dealt with a massive data set that provided information of each tornado down to the time, day, month, and year, the challenge was looping through each day consecutively, then month, then year – which I have achieved by creating loops within loops. Once the days of the month are looped through, they automatically start from the beginning of the next month, and similarly, once the months of the year have been looped through, they automatically start from the beginning of the next year, and so on.

Another challenge was the timeframe the user could view the pattern change over. I did not want to eliminate the accuracy of mapping the tornadoes on a daily basis, and at the same time, I wanted to allow the user to view pattern changes on a monthly basis throughout the years as well. This could allow the user to analyze seasonal patterns throughout each year and comparatively analyze these patterns over years and decades. I have achieved this by mapping all tornadoes that have taken place throughout the chosen month in semi-transparent grayscale highlights, while the tornadoes that occur on the chosen day are drawn in bolder color scale, with colors and stroke widths symbolizing tornado intensity and width respectively.  To avoid confusion and enhance clarity, the exact date of the information being symbolized is shown in real-time – whether the animation is running or paused.

What I have found interesting upon exploration of the visualized data, is the patterns that I have noticed seasonally, geographically, and circumstantially. I have noticed that tornadoes often tend to occur in clusters, and significantly increase in number throughout the middle of the year (May, June, July), while significantly decreasing in the colder seasons – but not in intensity. I have also noticed that the South Eastern areas of the USA are almost consistently hit by tornadoes through all seasons of the year. The following screenshots exemplify this.

Month Analysis 2
Visualization of tornadoes throughout the months of the year 2004

18
An example of high intensity tornado, with statistics showing fatalities and injuries

The following screenshots show the visualization of tornadoes with the different background maps provided:

13 2

The google drive link that includes the csv file I worked with, the script, and the video can be found here.

 

 

 

 

Comments are closed.