Typhoons Exploratory Data Analysis

We scraped data of 3,102 typhoons and tropical depressions in the Pacific from 1949 to 2016 from Weather Underground (https://www.wunderground.com/). The dataset includes latitudes, longitudes, wind speeds, and pressures. In this article, we explored trends in the number of typhoons (and tropical depressions), latitudes, longitudes, and wind speeds over time. From here on, we refer to both typhoons and tropical depressions as “typhoon.”

Number of typhoons

We show below the number of typhoons recorded by Weather Underground from 1949 to 2016. There is a sharp drop in the number of typhoons recorded since 2000. Can this be due to real natural events or are these anomalies in the data collection?

typhoon counts

We visualized below the total no. of recorded typhoons from 1949 to 2016 for each month. Most typhoons were recorded in August, then September and October.

typh_count_mont

Wind Speeds

Below shows the distribution of wind speeds. Most wind speeds recorded were at 25 mph. The number of typhoons that reach higher wind speeds quickly taper off.

dist_wind

The ranges of wind speeds seem to expand going from the first half of the year to the second half of the year.

wind_month

From the data, wind speeds seem to increase since 2009 onwards. Is this due to data gathering changes like better equipment or is this an actual phenomenon?

wind_dist

Maximum wind speeds in the data peaked in the 1960s and in the late 2010s.

max_wind_speed

Average recorded wind speeds in the dataset peaked post-2010. Is this a real phenomenon?

avg_wind_speed

Longitude

Longitudes (East-West) on the average are the same over twelve months. February has the most squished distribution, while October has the widest distribution.

long_month

Longitudes seem to follow a uniform behavior until 2009 and onwards. Why is this the case?

long_dist

There is no discernible trend in average longitude over time.

avg_long

Latitude

Looking at the figure below, we find a bell-shaped curve in the distribution of latitudes over 12 months. Typhoons in the first quarter have latitudes at around 10 degrees then ramping up in the following months up to a median of around 25 degrees by August. From August to December, the distribution of latitudes decline back to around 10 degrees.

lat_month

Visually, we find that March has the lowest median Latitude. We visualize below the trends in the distribution of latitudes from the 1950s to 2016 for the month of March. We see a noticeable downward trend in latitudes of Typhoons. If this is not an anomaly in the data then why is this so?

lat_march

Looking at the distribution of latitudes over time, we find that from 2009 onwards, the distributions are noticeably more squished. We go back to the same questions we raised above.

lat_dist

The average latitude seems to follow a downward trend from the 1950s to 2016. Might it be that typhoons are on the average moving lower towards the Equator?

ave_lat_over_time

Summary

We explored typhoons data scraped from Weather Underground from 1949 to 2016. We notice some trends such as:

  • The declining number of typhoons in recent years vs past decades
  • Most typhoons occurred in August
  • Most frequent wind speed is around 25 mph, faster winds are rare
  • Maximum wind speeds peaked in the 1960s and late 2000s
  • Average wind speeds peaked in the late 2000s
  • Longitude distribution got squished in the late 2000s
  • Typhoon latitudes follow a distinct seasonality with typhoons in August moving in much higher latitudes and typhoons in March moving in much lower latitudes
  • Typhoons seem to move to lower latitudes over the years

However, we have to note that we are uncertain about the quality of data scraped from Weather Underground. It might also be the case that these trends are the result of anomalous data. It is recommended to validate the data or even look for other datasets such as those collected by PAGASA.

Acknowledgments

I would like to acknowledge Sashmir Yap and Jayson Yodico whom I collaborated with to mine and organize data, and finally develop a machine learning model for typhoon path prediction.

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Create a website or blog at WordPress.com

Up ↑

%d bloggers like this: