Smoothing of timeSeries data using convolution filters

How to smooth a TimeSeries using a convolution filter kernel from convolution and convolve function.

import matplotlib.pyplot as plt

from astropy.convolution import Box1DKernel, convolve

from sunpy.timeseries import TimeSeries

Let’s first create a TimeSeries from sample data

goes_lc = TimeSeries(GOES_XRS_TIMESERIES).truncate('2011/06/07 06:10', '2011/06/07 07:00')

Now we will extract data values from the TimeSeries and apply a BoxCar filter to get smooth data. Boxcar smoothing is equivalent to taking our signal and using it to make a new signal where each element is the average of w adjacent elements. Here we will use AstroPy’s convolve function with a “boxcar” kernel of width w = 10.

goes_lc = goes_lc.add_column(
    convolve(goes_lc.quantity('xrsa'), kernel=Box1DKernel(50))

Plotting original and smoothed timeseries

plt.ylabel("Flux (Wm$^{-2}$")
plt.title('Smoothing of Time Series')
plt.plot(goes_lc.quantity('xrsa'), label='original')
plt.plot(goes_lc.quantity('xrsa_smoothed'), label='smoothed')
Smoothing of Time Series

Total running time of the script: ( 0 minutes 0.524 seconds)

Gallery generated by Sphinx-Gallery