Typhoon Data Visualization and Exploratory Analysis

Posted on Nov 21, 2016
The skills the author demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.

Typhoon Visualization Through the Years

Nearly 1/3 of all tropical storms occur in the Western Pacific basin. Prior to the electric telegraph, there was virtually no method of forewarning a typhoon in time before landfall. The advent of aircraft reconnaissance and theย  introduction of radarย  around 1950 helped create early detection systems for typhoons, and with satellites imaging there is a whole bevy of data that can be gleamed right from the early formation of a tropical storm.

Currently, scientists are able to track the path of typhoons and hurricanes, both simply different names for a high intensity tropic storm, but it remains a challenge to track their intensity. A powerful Super typhoon or Category 5 Hurricane with major destructive force can dissipate into a limping tropical storm of Category 2.ย  Likewise, a tame 45 mph storm might suddenly gather momentum and unleash havoc on landfall. Further areas of interest include predicting typhoon formation and even number of typhoons within a given time period.

Because the potential economic damages, interference with transportation and potential of life, predicting typhoons is a means to prevent such damage, or at the very least brace for them. With the advent of improved computing performance and automated collection of data, much of the improvement of prediction relies on techniques such as statistical modeling and machine learning.

In our exploratory analysis of typhoons, we do a cursory examination of the best track fit data provided by the Japanese Meteorological Society.ย  The goal is to provide a visualization and a possible blueprint for further in depth analysis.

 

The Data Set

The Dataset, acquired from here, consists of a header with information such as name, international ID, and number of observations in 6 hour intervals. The observations consists of variables including location in latitude and longitude, Pressure of the center, and Maximum Sustained Wind Speed for observations 35 kt or greater. Moreover, the dataset includes information of typhoon radii and speeds corresponding to the radii.ย  As of August 30th we have 62,771 observations ofย  1,712 tropical storms, including typhoons, in the Western Pacific quadrant. Note that prior to 1977 the observations only include ID, location, and Pressure. Lastly, from the data we can calculate time elapsed and approximate the distance traveled by using the Haversine distance between coordinates.

 

Using ggplot2, we examine the typhoon dataset on three levels: typhoon within a given year, typhoons over the years, and typhoons over seasons.

Jan to Aug 2016

First, we plot the behavior of typhoons from Jan to Aug 2016:Typhoon Data Visualization and Exploratory Analysis

Typhoon Data Visualization and Exploratory Analysis

 

Though nice to visualize the paths taken, and the pressure at each data point, it isn't very clear to what's going on.

Wind Speed and Pressure

We plot the corresponding Maximum Sustained Wind Speed and Pressure of the 12 tropical storms (click to enlarge):

screen-shot-2016-10-21-at-5-53-38-pm screen-shot-2016-10-21-at-5-54-03-pm

 

Note the pressure drop off and corresponding maximum sustained windspeed increase for the typhoon Nepartak. Wind flows from high to low pressure, so intuitively pressure and wind speed should have an inverse relationship.

 

1951-2016

Let's go further and look at the aggregate data from 1951 - 2016:

Here we work with four variables and plot them against each other: Pressure of the center, Maximum Sustained Wind Speed, Time Elapsed and Distance Traveled (click to enlarge).

screen-shot-2016-10-18-at-2-35-55-pmscreen-shot-2016-10-18-at-2-37-51-pmscreen-shot-2016-10-18-at-2-35-16-pm

screen-shot-2016-10-18-at-2-36-06-pmscreen-shot-2016-10-18-at-2-37-34-pmscreen-shot-2016-10-18-at-2-35-39-pm

in Pressure vs Time Elapsed and MSWS vs Time Elapsed, the scatterplot doesn't reveal much, though again we can see they both have somewhat an inverse relationship. For Pressure vs Distance Traveled and MSWS vs Distance Traveled, though we see that a majority of observations fall between 7500 - 15000 km, again, the plot doesn't reveal much. However, note the strange vertical gap around 11500 km. This indicates a boundary of some sort, and whether it's a result of the data collection or an actual trend could warrant further investigation.

Distribution & Correlation

The last two the plots reveal linear correlation: Average Pressure and Wind Speed are negatively correlated as expected, (I'd imagine with regression the best fit line will have a slope very close to one),ย  and Time Elapsed and Distance Traveled are positively correlated. However, interestingly enough, there seems to be a steeper slope for observations within 0 -50 hours, though the data points are relatively few. The jump is again due to the distribution of distance traveled. In fact, let's plot the distribution:

 

screen-shot-2016-10-18-at-1-07-26-pm

Visually, we can see the mean and median are around 9000 and 10000 km, respectively. Of course, a count of these along with standard deviation could be easily computed, but we won't be going further than the eyeball test for the time being.

 

And lastly, for typhoons over the years,

screen-shot-2016-10-17-at-3-42-03-am

There's seem to be a cyclical pattern going on. Perhaps for further time series analysis?

screen-shot-2016-10-21-at-6-22-02-pm

 

Now do typhoons depend on seasons? Here we do simple split of "seasons" as a time period of 3 calendar months:

screen-shot-2016-10-18-at-12-48-22-pmscreen-shot-2016-10-17-at-2-29-46-am

Seasons

From here it's clear that the summer and autumn seasons have a large spike in the number of typhoons.

Let's see if we can visualize the number of typhoons in a different way, by plotting start and end points:

First, the autumn season:

screen-shot-2016-10-17-at-3-15-32-amscreen-shot-2016-10-17-at-3-52-41-am

Here it's clear that typhoons form along a latitude approximately between 0 up toย  25 degrees latitude, not only in Autumn but in other seasons ( click to enlarge):

screen-shot-2016-10-17-at-3-50-54-am screen-shot-2016-10-17-at-3-51-50-am screen-shot-2016-10-17-at-3-27-26-am screen-shot-2016-10-17-at-3-28-06-am

Findings

And a reduction of number of typhoons also follows a reduction of range for Spring and Winter,ย  with a majority around 0 -10 degrees latitude, which means closer to the equator. Since the equator has warmer temperatures, perhaps this is another direction to follow.

Also, from above we can see two distinct clusters of end points, one westward, one northward, along with end points in between those two ( click to enlarge):

screen-shot-2016-10-17-at-3-18-47-amscreen-shot-2016-10-17-at-2-28-57-am

So it seems that after formation, typhoons tend to travel either westward or northward, and perhaps a third direction in between. Perhaps wind direction in the atmosphere is at play here?

Further Exploration:

 

We touched upon discovering an inverse linear relationship between wind speed and pressure, and also between distance and time elapsed, though it wasn't very clear what the relationship of the other variables are. Since weather is typically a complicated phenomenon, machine learning such as regression would be necessary to further explore features and to make better predictions. Perhaps we can even incorporate the other features in the dataset.

We noticed a significant increase in typhoons in Summer and Autumn along with a cyclical pattern of typhoon count over years. Using visualization, we noticed patterns in typhoon starting points and ending points. As mentioned, perhaps we can continue exploring with temperature gradient and wind direction data for a clearer picture.

 

 

 

 

About Author

Luke Chu

Luke is an aspiring data scientist with a background in Applied Mathematics, Mathematical Modeling, and Programming. He started learning data science due to his love of game and sport statistics and exposure from his prior job at Riot...
View all posts by Luke Chu >

Related Articles

Leave a Comment

No comments found.

View Posts by Categories


Our Recent Popular Posts


View Posts by Tags

#python #trainwithnycdsa 2019 2020 Revenue 3-points agriculture air quality airbnb airline alcohol Alex Baransky algorithm alumni Alumni Interview Alumni Reviews Alumni Spotlight alumni story Alumnus ames dataset ames housing dataset apartment rent API Application artist aws bank loans beautiful soup Best Bootcamp Best Data Science 2019 Best Data Science Bootcamp Best Data Science Bootcamp 2020 Best Ranked Big Data Book Launch Book-Signing bootcamp Bootcamp Alumni Bootcamp Prep boston safety Bundles cake recipe California Cancer Research capstone car price Career Career Day citibike classic cars classpass clustering Coding Course Demo Course Report covid 19 credit credit card crime frequency crops D3.js data data analysis Data Analyst data analytics data for tripadvisor reviews data science Data Science Academy Data Science Bootcamp Data science jobs Data Science Reviews Data Scientist Data Scientist Jobs data visualization database Deep Learning Demo Day Discount disney dplyr drug data e-commerce economy employee employee burnout employer networking environment feature engineering Finance Financial Data Science fitness studio Flask flight delay gbm Get Hired ggplot2 googleVis H20 Hadoop hallmark holiday movie happiness healthcare frauds higgs boson Hiring hiring partner events Hiring Partners hotels housing housing data housing predictions housing price hy-vee Income Industry Experts Injuries Instructor Blog Instructor Interview insurance italki Job Job Placement Jobs Jon Krohn JP Morgan Chase Kaggle Kickstarter las vegas airport lasso regression Lead Data Scienctist Lead Data Scientist leaflet league linear regression Logistic Regression machine learning Maps market matplotlib Medical Research Meet the team meetup methal health miami beach movie music Napoli NBA netflix Networking neural network Neural networks New Courses NHL nlp NYC NYC Data Science nyc data science academy NYC Open Data nyc property NYCDSA NYCDSA Alumni Online Online Bootcamp Online Training Open Data painter pandas Part-time performance phoenix pollutants Portfolio Development precision measurement prediction Prework Programming public safety PwC python Python Data Analysis python machine learning python scrapy python web scraping python webscraping Python Workshop R R Data Analysis R language R Programming R Shiny r studio R Visualization R Workshop R-bloggers random forest Ranking recommendation recommendation system regression Remote remote data science bootcamp Scrapy scrapy visualization seaborn seafood type Selenium sentiment analysis sentiment classification Shiny Shiny Dashboard Spark Special Special Summer Sports statistics streaming Student Interview Student Showcase SVM Switchup Tableau teachers team team performance TensorFlow Testimonial tf-idf Top Data Science Bootcamp Top manufacturing companies Transfers tweets twitter videos visualization wallstreet wallstreetbets web scraping Weekend Course What to expect whiskey whiskeyadvocate wildfire word cloud word2vec XGBoost yelp youtube trending ZORI