sigmoidsize_adv

sigmoidsize_adv performs a few measurements of sigmoids with some user intervention. The only input to the program is a keyword directory containing any number of sigmoid files (dir). First, the program sets the scale for each observation. Note that XRT images in the catalog are already normalized by exposure time, but that does not reset the EXPTIME header in the fits file. Instead, you must use keyword e_etim, which is in microseconds. For the maximum value the program just finds the 99.9 percentile in the image using cgPercentiles. Then to set the minimum scale value the program first looks for edges in the image using the edge_dog function in idl with radius1 = 3.6 pixels and radius2 = 15. pixels for 1 threshold value. Then use the edge_dog solution to create a mask for AR/bright regions in the image. If edge_dog solution does not detect any AR/bright regions the program does not apply a mask. In the masked image we interatively solve for the median background level. The program stops searching for the minimum after 95% of the pixels are within 2 sigma of the median or 6 iterations. Generally, the program finds the minimum level in less than three loops.

Next, the program will ask the user to perform a series of tasks. First, it asks whether we should analyze the file in the command line. Enter 1 if we should skip the file or enter nothing if we should analyze the file. Only enter 1 if the file does not contain a sigmoid you can analyze. If you enter 1 the program will give null values for ever parameters and go to the next file. If you enter nothing the program will ask you to click a number of places on the sigmoid.

Then you will select the long axis of the sigmoid. We define the long axis as the longest point on the sigmoid by eye using the following procedure. First, click the trailing end of the long axis as prompted in the command line. Once you click the trailing end, you should notice a L appears where you clicked. The command line prompt change will change to Click leading sigmoid long axis. Again click the leading edge of the long axis where a L will appear.

After that, we then compute the short axis length of the sigmoid at the trailing and leading ends. Similar to the long axis, the short axes are defined by inspection. The program will ask you in order to click the southern trailing short axis, northern trailing short axis, southern leading short axis, and northern leading short axis. Everytime you click a new appears.

Again the program asks for more human input. This time it is to put a box around the sigmoid. The box is important because it helps define the bounding box for the background in the Full Width Half Maximum (FWHM) calculation, and it helps the program flare_cme_sigcat locate nearby flares.

Finally, the last of the human input. The program asks for the core of the sigmoid to determine the FWHM. Click first point in the trailing core of the sigmoid, and the second point in the leading core. Those two point set the x boundary from which to calculate the FWHM. Once you select the leading edge of the sigmoid core, two new windows will pop up in the upper left. The farthest to the left is an image of the sigmoid rotated by the slope of the line selected in the sigmoid core. Highlight in the image is the region used to calculate the sigmoids FWHM. If you notice any irregularities due to structures not a part of the sigmoid reclassify the sigmoid using a new box that exclues the structure.

The calculation of the FWHM is a little tricky, so I will described it in some detail. The first problem is how does one compress a 2D sigmoid into a 1D shape. My solution is to rotate the sigmoid about its core and sum pixels along the x-axis for a given rotated y position (you can see the idea in the rotated image). After you sum the pixels you must now divide by the total pixels used in the sum, since FWHM is over a small window of the image I sum a rotated mask created from the sigmoid selection of the line core and the bounding box. I then divide the total counts by the sum of the masked bounding box. After summation and normalization, I find define the peak value (stored as hght) in the rotated y direction of the sigmoid profile. Then I break the 1D sigmoid profile into two halves. I then find the value closest to the half maximum for each half. If a half contains more than 1 value close to the half maximum I use the mean of the values to set the width.

In order to get a consistent FWHM, the program needs to subtract a background flux. The background finding algorithm uses the edge_dog AR mask of the sigmoid to mask out the bright regions and derive a first median background level. Then the program begins a loop to further restrict the background flux. First, it creates a new mask which includes pixels not in the AR mask and pixels less than three sigma (RMS) from the median value. From the new mask the program computes a new median background level and standard deviation. The program exits with 95 percent of the “background” pixel are contained within two sigma or after 6 attempt to find the background value. The program usually converges in 3 iterations. Finally, to subtract from the image in physical units and derive the background for the FWHM the pixel size is converted to arcsec.

After all computation are recorded, the program does some classification. First, the program asks for the sigmoid ID number. It will show you a best guess identification number based on previous identification, but you should verify it is correct before entering. In addtion, it will return an AR number based on the nearest AR from the HEK +/-7days. Again you should not blindly follow the returned number and verify by eye especially if many AR are in the region.

All the information is saved in an IDL save file in same directory defined in the dir keyword. The stricture of the save file has the following format:

sigdat_mod={sig_id:’‘, –> User defined Sigmoid ID

NOAA_id:0, –> NOAA ID of closest AR. filename:’‘, –> Fits file name (without directory) of the file used for Sigmoid classification date:’‘, –> Date of observation from the fits header size:0.0, –> long axis size of the sigmoid in arcseconds sizea:0.0, –> trailing short axis size of the sigmoid in arcseconds. sizeb:0.0, –> leading short axis size of the sigmoid in arcseconds

[early versions duplicated the short axis length](https://github.com/jprchlik/sigmoid_selector/commit/bab8b4ba6db4827b8896cbe12c61558b4bcafa5d#diff-3f0fc4ecafcfd45723c346c14af5de2d).

aspect_ratio:0.0, –> size/((sizea+sizeb)/2.) cx:0.0, –> Center of sigmoid in arcseconds from automatic region finding (BETA feature) cy:0.0, –> Center of sigmoid in arcseconds from automatic region finding (BETA feature) peri:0.0, –> Perimeter distance in arcseconds from automatic region finding (BETA feature) area:0.0, –> Area in square arcseconds from automatic region finding (BETA feature) roi:OBJ_NEW(‘IDLanROI’), –> Region object from automatic region finding (BETA feature) bkgd:0.0, –> Background flux in ADU/s/arcsec^2 fwhm:0.0, –> Half width of the sigmoid core in arcseconds hght:0.0, –> Peak flux of the sigmoid in ADU/s/arcsec^2 bboxx:fltarr(5), –> x-coordinates which make a bounding box for the observation in pixels bboxy:fltarr(5), –> y-coordinates which make a bounding box for the observation in pixels fwlin1:fltarr(2), –> x,y coordinates of the trailing end of the sigmoid core in pixels fwlin2:fltarr(2), –> x,y coordiantes of the leading end of the sigmoid core in pixels longx1:0.0, –> Trailing Long Axis X coordinate in Pixels longy1:0.0, –> Leading Long Axis X coordinate in Pixels longx2:0.0, –> Trailing Long Axis Y coordinate in Pixels longy2:0.0, –> Leading Long Axis Y coordinate in Pixels shrtx1a:0.0, –> X coordinate of Lower Trailing Sigmoid short axis in pixels shrty1a:0.0, –> X coordinate of Upper Trailing Sigmoid short axis in pixels shrtx2a:0.0, –> X coordinate of Lower Leading Sigmoid short axis in pixels shrty2a:0.0, –> X coordinate of Upper Leading Sigmoid short axis in pixels shrtx1b:0.0, –> Y coordinate of Lower Trailing Sigmoid short axis in pixels shrty1b:0.0, –> Y coordinate of Upper Trailing Sigmoid short axis in pixels shrtx2b:0.0, –> Y coordinate of Lower Leading Sigmoid short axis in pixels