Brian Dennis
Last modified 18 May 2002
This is a beginner's guide to RHESSI imaging. It assumes that you have already set up your computer to run IDL (Version 5.3.1 or later, preferably 5.4 or 5.5) with the Solar Software (SSW) tree and with access to the RHESSI flight data (see Software Installation Instructions and Accessing HESSI Data). It will take you through demonstrations using the Graphical User Interface (GUI) to the RHESSI software. Another guide to RHESSI imaging using the IDL command-line interface has been written by Chris Johns-Krull.
The objective of this document is to demonstrate how to obtain images for different time and energy intervals during a flare. Imaging spectroscopy is not covered in this demonstration; further analysis is required to obtain spectra for different areas within an image.
The GUI is a user-friendly way to obtain light curves, images, and spectra. This demonstration just covers light curves and images. Go to Spectroscopy - First Steps for a demonstration of how to obtain spectra. You can find more details on all of the RHESSI software from the RHESSI Data and Software Center.
First, let's select a flare using the Graphical User Interface (GUI). Enter IDL so that it uses SSW and has access to the RHESSI flight data. You should see the following standard prompt in the IDL window:
IDL>
To enter the GUI, simply type the following after this prompt:
hessi
You should get a window that looks like this:
We begin by choosing the observing time interval containing the flare of interest. For this demonstration, we have chosen a flare that occurred on February 20 at 11:00 UT. The selected interval includes time before and after the flare to allow good background spectra during the flare to be obtained.
On the command bar at the top of the GUI window, pull down "File" and choose
Select Observation Time Interval
A new window will open similar to the one shown below. It allows you to select data by date and time, by looking at RHESSI's flare list, by picking a flare by number, or by looking at the Observing Summary data and picking a time interval from the overall light curve.
We will do it by specifying the start and end times. You can type them directly into the "Start..." and "End..." windows or you can click the "Start" button and enter the following date and time in the window that opens up:
Year 2002, Month February, Day 20, Hour 11, Minute 0, Second 0
Click the "READY" button and the following should appear in the "Start" location of the Observation Time Selection Window:
20-Feb-2002 11:00:00.000
You can enter the end time in the same way by pressing the "End" button or you can click on the "Duration (s)" box and edit its contents by typing "3600" over whatever is already there. You must then press the "Enter" key and the following end date and time should appear in the box next to the "End" button:
20-Feb-2002 11:20:00.000
The Observation Time Interval Selection window should now look like the following:
Note that flare number 200203 occurs in this time interval and you could have used the "Flare Selection..." button to select it instead of specifying the start and end time.
Click on the "Plot Observing Summary Data" button and you will see the following quicklook light curves appear in the Main window:
If you press the little downward-pointing triangle to the right of the "Data type" box and select either "Count Rate"or "Ephemeris" or Modulation Variance," and press the "Plot Observing Summary Data" button again you will get a plot of the corresponding parameters for the specified observation time interval. (The other options in this box are not yet implemented.) Press the "Change" button in the "Show Flags" box and you can select the flags to display on any of the plots. The selections are SAA (South Atlantic Anomaly), Night, Flare, Decimation, Attenuator, and Fast Rate.
Once the correct Start, Duration, and End dates and times appear in the respective boxes, click the "Accept and Close" button to exit back to the opening GUI window.
Now let's look at a light curve to decide when we want to make images. Back on the main GUI window, use the pull-down menu on the top to select:
File ------- Retrieve/Process Data ---------- Lightcurve...
and a new window labeled LIGHTCURVES should open up looking something like the following:
Note that the defaults are to plot the light curves for the full observation time interval with 0.1-s time bins and one energy band from 6 - 50 keV. To avoid any problems with the modulation of the flux affecting the light curve, we want to make the plot with time bins just one rotation period long. For this flare, the spin period was 4.3329 s so we need to enter this number into the "Time Resolution (s)" box in this window. (See the separate instructions for determining the spin period for a specific time.)
We want to select the 9 standard energy bins defined by the following energies: 3, 6, 12, 25, 50, 300, 800, 7000, 20000 keV. To do this, press the "Change" button in the Energy Bands box to produce the following "Select Energy Interval(s)" window:
There are various ways to enter energy intervals in this window but for now we will just use the "All Standard" button near the bottom right of this window. Press this button and you will see 8 highlighted energy interval appear in the middle box of this window. Click on the "Replace List" button in the middle right of the window next to the highlighted list that just appeared and the "List of current intervals:" at the top of the window should now show "# intervals = 9." The box under this should show "Interval 1, 3.000 to 6.000." At this stage the "Select Energy Interval(s)" window should look like this:
finally, click on the "Accept" button and the "LIGHTCURVES" window should now show the 9 energy bands as well.
All detector front segments are selected and summed together by default. We will accept all of the other default settings so, once you have done this, press the "Plot Lightcurve" button and after a few minutes you should see the following light curves appear in the main GUI window, one for each energy bin:
Click on the "Close" button in the LIGHTCURVES window.
I prefer to see the light curves plotted on a log scale to show up the higher energies better. To do that, click on the "Plot_Control" pull -down menu and select XY-Plot display Options and you will get the following window:
There are many ways in this window to control the way the light curve is displayed but for now, click on the "X axis Exact" box and on the Y axis: Log" box. Also, Click on the little downward pointing triangle to the right of the "Legend location:" box and select "Upper right." This will move the legend so that it does not appear over the peak of the flare. Press the "Accept and Close" button, choose the "Apply to this plot only" option, and the light-curve should now appear as follows in the Main Window:
Click on the "Accept and Close " button to get back to the GUI "Main Window."
Finally, we are ready to make images. Use the pull-down menu on the top of the GUI "Main Window" to select:
File ------- Retrieve/Process Data ---------- Image...
and a new window labeled IMAGING should open up looking something like the following:
By default, a 4-s interval at the peak of the flare has already been selected. But we want to choose an interval at the peak of the flare that is an integer number of rotations periods to be sure to sample all of the Fourier components uniformly. To do this, click on the "Change" button in the "Image Time Interval" box. In the "Select Time Interval(s)" window that appears, click on the "Edit intervals..." button near the top right of this window. Set the start and end times to the following:
Interval 1, 20-Feb-2002 11:06:05.670 to 11:06:27.370
Note that this is exactly 5 rotations and spans the time of the highest counting rates in the highest energies. Click on the "Accept" button and note that the correct start and end times appear in the IMAGING window.
The next thing to do is to find out roughly whereabouts the flare was on the Sun. Ultimately, this will be automatically determined during quicklook processing for flares that are identified from their modulation signature. For now, this must be done by the user as described in separate instructions for finding the flare location from the RHESSI data. We will use the results of following those instructions to obtain the X and Y coordinates of the center of the flare. Click on the "Change button in the IMAGING window in the box with the Pixel Size, Image Dimensions, etc. In the "Imaging Size and Location Options" window that opens up, type in the following numbers for the "Offset of map center from Sun center (arcsec)," being sure to press the "Enter" key after each number:
X: 910, Y: 270
Change the pixel size to 4x4 arcseconds and the image dimension, "Image Dim (pixels):" to 64x64.
This window should now look like this:
Click on the "Accept" button to get back to the IMAGING window and note that the new "Offset of Map Center from Sun center (arcsec) appear.
The last thing we need to do before making our first high resolution image is to decide which collimators we want to use. For now, the two finest collimators, #1 and 2, should not be used for imaging since their relative front-to-rear phases have not been determined. Thus, we will stick with collimators 3 - 8. Collimator 9 will not contribute much to the image since the source is compact compared to its 3-arcminute resolution. Click on the "Change..." button in the IMAGING window if these are not already the collimator numbers that appear bolded. The following window should appear:
Make the necessary adjustments to make this window appear as above and then click on the "Accept" button.
Finally, we are ready to make our first high resolution image using the default "Back Projection" reconstruction process. Click on the "Make and Display Image(s)" button and after a short while depending on the speed of your computer, you should see the following image in the GUI Main Window:
Note the bright source near the center of the square indicating that we have chosen the correct offset from Sun center. The red line shows the solar disk.
The concentric rings around the extended bright patch are the side lobes always present when you use "back projection.". They can be removed by changing the "Image Algorithm" to "Clean" in the IMAGING window that should still be open. Don't expect Clean to improve the resolution; it only removes side lobes. Click on the "Make and Display Image(s)" button again and you will see the HSI CLEAN window open with a Progress Bar. It's neat to actually watch as CLEAN does its thing and removes the side lobes to give the cleaned image in the GUI Main Window.
You can make this a color image by using the "Plot_Control" pull-down menu and selecting "Image Colors." Scroll down in the Select Color Table of the "Plotman colors" window that opens up and highlight "4 -BLUE/GREEN/RED/YELLOW" as follows:
You should see the following colored "Cleaned" image in the GUI Main Window:
You can play with the color table or select a different one of your choice to get the color scale that looks best to you. Click on the "Accept" button once you like the image's appearance.
In this image you can begin to see the double-source nature of this flare that has made it so interesting to analyze. You can still see some arcs of circles centered on the location of the spin axis that indicate discrepancies in the flat-fielding process and in the point-spread functions that are used in the CLEAN process.
This completes the demonstration of the basic technique for obtaining RHESSI images. Now that you have been through the basic steps, you can try to refine the image by choosing different detectors, changing the pixel size, using different time intervals, and using one of the other image reconstruction techniques - a Maximum Entropy Method (MEM Sato or MEM VIS), Pixon, or Forward Fitting. You can also make images for different energy ranges and multiple time intervals and, when you get really proficient, you can make a movie file that you can play back in the GUI Main Window.
Remember that all of the plots that you made during any GUI session are saved for the duration of the session. They can be accessed from the "Window_Control" pull-down menu in the GUI Main Window. You can also save any plot in one of several different formats (ps, png, tiff, or jpg) or print it out, provided that you have set up access to a suitable printer for IDL to use. Do this from the "File" pull-down menu on the GUI Main Window" and select the appropriate option. Note that the print option crates a postscript file in the IDL working directory "plotman_print.ps" by default, and then sends it to the printer that you select. As a last resort, you can always select the window containing the plot you want to print and press the Alt and Print Screen keys simultaneously. This puts an image of the selected window in your Clipboard and allows you to paste it into any program such as Word, PowerPoint, Irfanview, etc. that can handle such files. From there you can print the image or save it in the format of your choice.