Kilauea 2018 earthquakes - Visualization with Bokeh

The 2018 eruption of Kilauea volcano in Hawaii was accompanied by thousands of small earthquakes over the course of the eruption. This seismicity exhibited an interesting pattern: cycles of increasing earthquake frequency, each ending with a larger magnitude explosive event. Over the course of the 2+ month eruption, there were over 50 of these cycles.

The Bokeh plotting library provides a lot of great tools to look at a dataset like this, where it might be useful to zoom in on certain cycles or look only at a particular date. Here, we will start with a simple plot, using only circles and line segments.

# Plotting
import bokeh
from bokeh.plotting import figure, output_file, show
from bokeh.io import output_notebook
from bokeh.models import Range1d

# Data analysis
import pandas as pd
from datetime import datetime

USGS earthquake data

The earthquake event data was downloaded from the USGS Earthquake Catalog. Before we could use it for this plotting exercise, it needed to be cleaned up a bit. After combining the separate files, dropping uneeded columns, and sorting by increasing date, the data is ready to be loaded in.

# Read in the data for the earthquakes
events = pd.read_csv('/home/nicole/earthquakes/kilauea/data/all_events_M15.csv')
events.head()

	DTtime	depth	mag	magType
0	2018-04-28 11:29:50.700	0.27	2.08	ml
1	2018-04-28 23:09:53.430	0.27	2.12	ml
2	2018-04-29 05:46:18.950	0.23	1.90	ml
3	2018-04-29 09:33:02.850	1.39	1.95	ml
4	2018-05-01 06:34:32.670	1.81	1.78	md

# Read in the data for the explosions
explode = pd.read_csv('/home/nicole/earthquakes/kilauea/data/all_explode_M4.csv')
explode.head()

	DTtime	depth	mag	magType
0	2018-05-17 04:15:30.350	0.01	5.0	mw
1	2018-05-17 14:04:10.700	0.01	5.0	mw
2	2018-05-19 09:58:33.210	0.01	5.1	mw
3	2018-05-20 01:58:13.320	0.01	4.9	mw
4	2018-05-20 21:50:07.310	0.01	4.9	mw

Converting to a datetime object

The “DTtime” column is not currently in a datetime format; plotting will be easier if we convert the column to a numpy.datetime64 object.

# Convert "DTtime" to a datetime object

print("Time column format: ", events['DTtime'].dtype)
events['DTtime'] = pd.to_datetime(events['DTtime'])
explode['DTtime'] = pd.to_datetime(explode['DTtime'])
print("Time column format (converted): ", events['DTtime'].dtype)

Time column format:  object
Time column format (converted):  datetime64[ns]

Make the plot!

We are ready to make our Bokeh plot. The earthquake will be plotted as circles, with time on the x-axis and earthquake magnitude on the y-axis. To better mark the time of the explosive event, they will be plotted as vertical lines with a height equal to the magnitude.

To display inline a Jupyter notebook (which I am using), we use the output_notebook() function.

# Display plot in the notebook
output_notebook()

Loading BokehJS ...

# Make vertical line coordinates for each explosion

# x0, x1 = time of explosion
# y0 = 1.5, y1 = magnitude of explosion

ex_time = explode['DTtime']
ystart= [1.8] * len(ex_time)
yend = explode['mag']

# Make the plot
output_file('kilauea2018_cycles.html', title='Kilauea Summit Earthquakes')

p = figure(plot_width=1000, plot_height=600,
           x_axis_type='datetime',
           x_axis_label = 'Date',
           y_axis_label = 'Magnitude',
           y_range = Range1d(1, 7),
           title = 'Kilauea summit earthquakes')

p.segment(x0=ex_time,y0=ystart,x1=ex_time,y1=yend,
         line_width=1, color='gray', alpha=0.4)

p.circle(x=events['DTtime'], y=events['mag'],
         size=1.5, color='darkblue', alpha=0.3,
        legend='earthquakes (Ml or Md)')

p.asterisk(x=explode['DTtime'], y=explode['mag'],
          size=10, color='gray', alpha=0.8,
          legend='summit collapse event (Mw equivalent)')

p.legend.location = "top_left"
p.legend.orientation = "horizontal"
p.legend.background_fill_alpha = 0

# To view the plot inline, uncomment the following line
#show(p)

In order to be able to interact with the figure we created above, we need to save the html of the file and then place that code in the static directory of the hugo site. I created a directory for my Bokeh plots and subdirectories for different topics.

First, import the necessary methods from the Bokeh library and create the html for the plot. Then write out the html to a file, close the file, and move it into the appropriate directory. When the Hugo site is built, finding the image will be as easy as clicking on the link below.

from bokeh.embed import file_html
from bokeh.resources import CDN

html = file_html(p, CDN, "Kilauea 2018 earthquakes")

kilauea2018=open("kilauea2018.html",mode="w",encoding="utf-8")
kilauea2018.write(html)
kilauea2018.close()

Interactive plot!

Check out the interactive image!