Spectrogram Display and Analysis Software
Reading and Plotting the Dynamic Spectra
RSTO data can be analyzed in IDL in Python or with our LabView GUI. In the FITS archive you will find a directory listing of YYYYMMDD, please note that the current day fits files are found in a folder named 'realtime'. In each folder you will find a 'fts' directory, containing the sub directories low_freq (10-100 MHz), mid_freq (100-200 MHz), high_freq (200 - 400 MHz). Each directory contains fits files named with the follwing key BIR_YYYYMMDD_HHMMSS_XX.fit. The first three characters are the site identifier (Birr Castle in this case), followed by the year, month, day and hour, minute, second labels. The XX numerals are an identifier for our own purposes and have no physical meaning. Please download the data of your choice.
Figure 1. Default plot for data1 and data2. Note the default does not display the frequency axis,
or fill the array with available plot space. The radio bursts are not very discernable.
Insert the two spectrograms and time arrays into respective master arrays in order to view the two as a whole spectrogram, and replot
Figure 2. Default plot for data1 and data2 stacked together into single spectrogram
The plotting procedure can use the usual keyowrds for better data display. Choose preferred colour chart, force the spectrogram to fill the window, and add appropriate axes labels and names. Choosing appropriate time and frequency ranges can allow a more detailed view of the burst structure.
IDL> spectro_plot,data,time,freq,/xs,/ys,charsize=1.5,title='RSTO low CALLISTO receiver',$
To save the plot in the current graphics device use the simple command
The above should produce a jpeg with the following dynamic spectra in Figure 3.
Figure 3. Dynamic spectra from RSTO 'low' CALLISTO receiver, covering the frequency range of 10-100 MHz.
This is a type II event associated with and X1.4 class flare on 2011 Sepetember 22, at ~10:30 UT .
Dynamic Spectra Background Subtraction
Radio frequency measurements are almost always subject to some level of radio frequency interference (RFI) or background noise. For example Figure 3 shows quite clearly a type II radio burst, however there is also other signals in the spectra such as FM commercial radio broadcast that can be seen between 90-100 MHz as well as continuous strong bands of interference such as the dark line at ~22 MHz. RFI may also be discrete in time, having small duty cycle such as the 'blips' at ~70 MHz between 10:39:20 UT - 10:40:20 UT. There are many software packages available within SolarSoft that employ various methods of background subtraction. More sophisticated technqiues may be required to remove short duty cycle RFI or other unwanted discrete signals in the dynamic spectra. The following is a guide to some simple techniques of background subtraction.
A standard background subtraction function is to use constbacksub.pro, applied as follows
IDL> data_bg = constbacksub(data,/auto)
This function firstly computes the average of each frequency channel in time and subtracts this from the original spectrogram. It then computes the standard deviation of each timestep. The timesteps containing a radioburst should have the largest standard deviation, while those without a burst will have small standard deviation. If the '/auto' keyword is selected, it choose the lowest 5% of the these standard deviation time steps, averages them over time, and creates a background with this average that is the size of the input array. It then subtracts this background from the original array.
For the result of the process plot the spectrogram as above. Results are in Figure 4.
IDL> spectro_plot,data_bg,time,freq,/xs,/ys,charsize=1.5,ytitle='Frequency [MHz]',$
title='RSTO low CALLISTO receiver',yr=[100,10],ytickv=[100,80,60,40,20,10],yticks=5,yminor=4,$
Figure 4. Dynamic spectra from RSTO 'low' CALLISTO receiver, covering the frequency range of 10-100 MHz.
This is the same dynamic spectra as above, but with a standard background subtraction to remove some of the RFI.
The process works quite well provided the RFI is contant over time. The commercial FM radio band has been eliminated, as has the RFI in channels below 40 MHz. As mentioned this background technique is innefective at removing short duty cycle RFI spikes in the data. The removal of such signals requires more sophisticated techniques. Another example of simple background subtraction can be found in Figure 3, showing a seperate type of radio activity from above.
Figure 5. Dynamic spectra from RSTO 'low' CALLISTO receiver, covering the frequency range of 10-100 MHz.
These plots show herrigbone structure from activity associated with the same event as the type II above. The
upper panel is the raw data, the lower panel shows the background subtracted data. The lower panel reveals
fainterburst with postive slopes moving towards lower frequencies, which are otherwise obscurred in the raw data.
constbacksub.pro is part of a number of similar background subtraction procedures which employ similar techniques such as glidbacksub.pro, the implementation of which can also be used interactively using the ragview GUI.
More information on similar techniques to those described above can also be found at ETH Zurich's software pages.
For any comments or queries regarding the information set out on this page please see the RSTO contacts page
We would appreciate if users of our data would include the following acknowledgment "Callisto data supplied courtesy of Trinity College Dublin".