The world’s leading publication for data science, AI, and ML professionals.

A Visualization of Decrease in Arctic Sea Ice Extent Over Past 40 Years

An animated data story made with Flourish…

Visualize Decrease in Arctic Sea Ice Extent with Flourish

An animated data story over past 40 years

Photo by Eva Blue on Unsplash
Photo by Eva Blue on Unsplash

Hi!

This article is primarily written to present my Udacity Visualization Nanodegree capstone project. Within this article I would like to show you how to make a dynamic radar plot with Flourish and create the presentation slides with audio & auto-play enabled!

Background Photo Source
Background Photo Source

Dataset & Clean-up

As required by the Udacity program, I am only able to choose a dataset from MakeoverMonday website. After carefully searching, I decided to work on the Arctic Sea Ice Extent dataset because I have always been interested in this beautiful world of ice. And I am curious to learn more about how fast the ice has been decreasing over the past few decades.

The data source is from National Snow & Ice Data Center. And it’s shared with everyone on the Makeover Monday website.

The dataset is quite simple with 12,751 rows and 2 columns. One column is for the yyyy-mm-dd form of date, and the other is for the numeric value of ice extent in unit of million square kilometres.

Compared to the original visualization on website, I want to improve it to be more attractive for readers and make sure people can feel the change over decades.

In order to make the dataset easier to visualize, I wrote some python code to process the dataset into the following format:

Cleaned Dataset (Image by Author)
Cleaned Dataset (Image by Author)

Essentially, the cleaning up code looks like the following:

import pandas as pd
import calendar #convert month number to month name
df = pd.read_csv('data.csv') #replace with your dataset name
df.Date = pd.to_datetime(df.Date) #Date column to datetime data type
df['month'] = df.Date.dt.month #create new column of month number
df['year'] = df.Date.dt.year #create new column of year 
df = df.drop('Date',1) 
#convert month number to month name
df['month'] = df['month'].apply(lambda x: calendar.month_name[x])
#group by year and month to calculate monthly average for each year
df = df.groupby(['year','month']).mean().reset_index()
#pivot the dataframe to make each month as a new column and year as index
df = df.pivot('year', columns='month')
df.head()

After above steps, you may find the dataset columns a bit messy. I manually delete the redundant columns & rows and only keep columns of year and month names.

Data Visualization with Flourish

The really nice part of data visualization on Flourish is that you can animate the visualization and upload an audio file that will auto-play along with the presentation.

Flourish provides so many visualization format with examples, for this project, I chose the Radar with filter.

Select Visualization Type (Image by Author)
Select Visualization Type (Image by Author)

Once you clicked in, you will see a button on top that allows you to switch between Preview and Data. To upload the dataset we just cleaned up, you need to click the Data option first. Then you may need to change the values for Name, Values and Filter according to the dataset columns.

Eventually, the Data page looks like the following for me:

Data Page (Image by Author)
Data Page (Image by Author)

Now, if we click back to the Preview, you may see Radar plots by each year. On the right side, there are many options for us to add backgroud image and adjust colors, word fonts etc.

Preview Page (Image by Author)
Preview Page (Image by Author)

In order to view the Radar plot change on a yearly basis instead of all of them on one page, you could change the Filter Control to the slider type under Controls on the right side bar.

Filter Control (Image by Author)
Filter Control (Image by Author)

The visualization may start from ‘All’ by default, once you slide the slider to the right side, it should start to go over each year and the Radar plot will change accordingly.

Background Photo Source
Background Photo Source

Data Story with Flourish

Finally, we can turn your previous visualization into a data story!

You could create a new story on the main page by clicking New Story:

Flourish Main Page (Image by Author)
Flourish Main Page (Image by Author)

Essentially, once you’re in the story page, you can either add a new slide from one of the existing visualizations or select a basic static slide. As for a basic slide, you can upload your own background image by replacing the image URL link.

Finally, if you are interested, you can upload your audio recording in mp3 format to the data story. You may need to practice a few times to make sure your audio presentation fits perfectly with the slides auto-play speed.

Click Enabled for Audio Upload (Image by Author)
Click Enabled for Audio Upload (Image by Author)

Limitations & Bias

In terms of existing limitations & bias in this project, I want to point out several as following:

  • Data Collection Bias: Only certain days of a month are recorded in the dataset so the average values of each month may not be accurate enough.
  • Data Processing Bias: Although there are no missing values for my selected years, there still may exsit some outliers without proper treatment.
  • Data Insights Bias: Currently we view the change qualitively and learn the decrease over past 40 years but ideally we should give a quantitatively calculation and publish more accurate estimation results.

Conclusion

As a conclusion, I am surprised about the speed of Arctic Sea ice extent decreasing. I have only seen what Arctic Sea looks like from documentary, but after seeing this visualization, I feel that the history is in front of us. As far as I learned, many of the icebergs have been on earth for thousands of years. However, suggestions said that before mid-century we could have a nearly ice-free Arctic in the summer.

Finally, here is the link to my capstone presentation with Flourish.

Thanks so much for your reading!

Data Citation

_Sea ice extent and area organized by year_ courtesy of the National Snow and Ice Data Center, University of Colorado, Boulder

Fetterer, F., K. Knowles, W. N. Meier, M. Savoie, and A. K. Windnagel. 2017, updated daily. Sea Ice Index, Version 3. Sea ice extent and area organized by year. Boulder, Colorado USA. NSIDC: National Snow and Ice Data Center. doi: https://doi.org/10.7265/N5K072F8


Related Articles