Coronal Loops Tracing

This example traces out the coronal loops in a FITS image using occult2. In this example we will use the settings and the data from Markus Aschwanden’s tutorial on his IDL implementation of the OCCULT2 algorithm, which can be found here. The aim of this example is to demonstrate that occult2 provides similar results compared to the IDL version.

import matplotlib.pyplot as plt
import numpy as np

from astropy import units as u
from import fits

import sunkit_image.trace as trace

We will be using to read the FITS file from the tutorial website and read in the header and data information.

hdu ="")[0]

# We can now make this into a ``. There is currently not an instrument specific
# class for the TRACE instrument.
trace_map =, hdu.header)

# We can now plot the map, of which we can see coronal loops.
TRACE 171 1998-05-19 22:21:43
INFO: Missing metadata for solar radius: assuming the standard radius of the photosphere. []
/home/docs/checkouts/ SunpyMetadataWarning: Missing metadata for observer: assuming Earth-based observer.
For frame 'heliographic_stonyhurst' the following metadata is missing: dsun_obs,hgln_obs,hglt_obs
For frame 'heliographic_carrington' the following metadata is missing: crln_obs,crlt_obs,dsun_obs

  obs_coord = self.observer_coordinate

<matplotlib.image.AxesImage object at 0x7f11440e9160>

Now the loop tracing will begin. We will use the same set of parameters as in the IDL tutorial. The lowpass filter boxcar filter size nsm1 is taken to be 3. The minimum radius of curvature at any point in the loop rmin is 30 pixels. The length of the smallest loop to be detected lmin is 25 pixels. The maximum number of structures to be examined nstruc is 1000. The number of extra points in the loop below noise level to terminate a loop tracing ngap is 0. The base flux and median flux ratio qthresh1 is 0.0. The noise threshold in the image with repect to median flux qthresh2 is 3.0 . For the meaning of these parameters please consult the OCCULT2 article.

loops = trace.occult2(, nsm1=3, rmin=30, lmin=25, nstruc=1000, ngap=0, qthresh1=0.0, qthresh2=3.0)

occult2 returns a list, each element of which is a detected loop. Each detected loop is stored as a list of x positions in image pixels, and a list of y positions in image pixels, of the pixels traced out by OCCULT2. Now plot all the detected loops on the original image, we convert the image pixels to to world coordinates to be plotted on the map.

fig = plt.figure()

ax = plt.subplot(projection=trace_map)


# We can now plot each loop in the list of loops. We plot these in world coordinates, converting them
# through the `pixel_to_world` functionality which converts the pixel coordinates to coordinates (in arcsec)
# on the `trace_map`.
for loop in loops:

    loop = np.array(loop)  # convert to array as easier to index `x` and `y` coordinates

    coord_loops = trace_map.pixel_to_world(loop[:, 0] * u.pixel, loop[:, 1] * u.pixel)

    ax.plot_coord(coord_loops, color="b")
TRACE 171 1998-05-19 22:21:43

Total running time of the script: ( 0 minutes 42.006 seconds)

Gallery generated by Sphinx-Gallery