# Spectra¶

Warning

This module is under development! It is being moved into its own repository.

## Spectrograms¶

SunPy currently supports reading dynamic spectra from e-Callisto instruments. The main class that is used for this is CallistoSpectrogram. SunPy also comes with an example image that shows a radio burst observed at Rosse Observatory (aka. BIR; Birr Castle, Co. Offaly, Ireland) that can be found in sunpy.data.sample.CALLISTO_SPECTRUM.

import matplotlib.pyplot as plt
import sunpy.spectra
import sunpy.data.sample
from sunpy.spectra.sources.callisto import CallistoSpectrogram
image.peek()


We now notice that there seems to be something interesting that has been cut off at the corner of the image, so we use the extend method to request more data from the server. It optionally takes the amount of minutes we want to request from the server (negative values mean we want to add data that was registered before our existing local data), if none are given it defaults to 15 minutes (the size of one e-Callisto file).

more = image.extend()
more.peek()


We will, for the purposes of this demonstration, continue working with the original image, though.

You can then perform automatic constant background subtraction by using the subtract_bg() method. The resulting image will be clipped at 0 using the min parameter of peek in order to avoid negative values.

from matplotlib import pyplot as plt
import sunpy
import sunpy.data.sample
from sunpy.spectra.sources.callisto import CallistoSpectrogram
nobg = image.subtract_bg()
nobg.peek(vmin=0)


If you want to see the background determined by the automatic subtraction, you can use the auto_const_bg() method and visualize the resulting data using pyplot.plot().:

plt.figure()
bg = image.auto_const_bg()
plt.plot(image.freq_axis, bg)
plt.xlabel("Frequency [MHz]")
plt.ylabel("Intensity")
plt.show()


Now let us say we want to isolate the interesting bit (which starts around 10:38) from the boring background; there is a method called in_interval() that allows us to take the part of an image that is within a specified interval. Leaving out the second argument it defaults to the end time of the file.

import matplotlib.pyplot as plt
import sunpy
import sunpy.data.sample
from sunpy.spectra.sources.callisto import CallistoSpectrogram
nobg = image.subtract_bg()
interesting = nobg.in_interval("06:27")
interesting.peek(vmin=0)


To get rid of the noise, we could also clip low intensities by setting vmin.

If we want more context, we can also join together different images into a large one in time (note that this does more than just concatenating the array and the axes – it also considers possible overlap or gaps).:

c1 = CallistoSpectrogram.read('BIR_20110922_101500_01.fit')

from sunpy.spectra.sources.callisto import CallistoSpectrogram