Tags:
create new tag
view all tags

Small Events Detection Deployment

See SmallEventsDeploymentResults for timing information on science cases 1 - 3.

Science Testing Document

Input

The Small Events algorithm requires a 3-D Fits data cube as input, which essentially holds a time sequence of 2-D images. The data cube can be of any size, subject to computer memory restrictions. The algorithm has been tested both on data cubes produced from simulated data, and data cubes produced from archived TRACE data. The input dataset itself remains unaltered by the processing. Optional input includes the size of the required macropixel and a text file containing Field of View restrictions if this is required. See the package README file or the Algorithm description webpage for further details.

Compilation and Execution:

  • gcc compilation:
    % gcc -g -o SmallEventsDetection -I/usr/local/include small_events_detection.c trace_variable.c dyn_mem_alloc_fns.c math_fns.c trace_ellipse.c structure_pattern.c event_struct.c small_events_ascii_output.c small_events_fits_output.c small_events_fits_lightcurve.c printerror.c ndArray/ndArray.c ndArray/where.c -L/usr/local/lib -lcfitsio -lm

A makefile (Makefile) is provided in the installation package to simplify building of the algorithm. The Makefile needs to be modified to point to the appropriate libraries and sub-directories, prior to building. Once this has been done, simply run make by typing:
% make force

which will create an executable called SmallEventsDetection

  • commandline execution:
    %
     ./SmallEventsDetection   <input filename><Size of Macropixel><FOV restrictions>

    , e.g. SmallEventsDetection datacube.fits 4 Y

  • input filename The name of the input 3D datacube fits file. The pathname should be included if the datafile is in a different directory to that of the executable, SmallEventsDetection. This is the only required input parameter.
  • Size of Macropixel the number of pixels which make up one side of a macropixel i.e if Size of Macropixel is set to 4, then a macropixel comprises 4x4 pixels. This input parameter is optional with a default value of 4 (i.e 4 pixel per macropixel side).
  • FOV restrictions This is another optional input parameter, and flags whether or not to restrict the field of view of the image. The default value is "No". If field of view restrictions are required a "YES" input parameter is needed here. (This can be input as Y,YES,y,yes, or any combination of these as long as the first letter is Y or y). If this option is selected the user needs to supply a file called "fov.txt" and this needs to reside in the directory where the executable is being run. If FOV restrictions are required, then the macropixel size option MUST also be included. (See README file for details).

  • Example Inputs are:

   ./SmallEventsDetection ../qs171ad.fits 
   ./SmallEventsDetection qs171ad.fits 4 
   ./SmallEventsDetection ~/data/qs171ad.fits 4 Yes 
   ./SmallEventsDetection ../../qs171ad.fits 4 YES 
   ./SmallEventsDetection ../qs171ad.fits 4 yES
   ./SmallEventsDetection ../qs171ad.fits 4 yESsireeBob 

  • AstroGrid workflow instructions:

Expected Output

If processing is successful, then two output files are produced. One is an ASCII file containing the output for each event in tabular form. The second is a fits file comprising three extensions. The main fits header unit is a copy of the input datacube image. The first extension is an ASCII table containing the same output as the ASCII file. The second extension contains the lightcurves for each event.

The tabular output for both the ascii and fits outputs are as follows:

_nr_
The event number in the output list.
_nstruc_
The number of macropixels comprising the event.
_tmin_
Time series array element number at which minimum occurs.
_tmax_
Time series array element number at which maximum occurs.
_xc_
X-coordinate of the center of ellipse enveloping event.
_yc_
X-coordinate of the center of ellipse enveloping event.
_len_
Length of major axis of ellipse.
_wid_
Width of minor axis of ellipse.
_al_
Orientation of ellipse major-axis to the image x-axis (in degrees).
_flux_
Total Flux contained within the event.
_fluxavg_
Average flux within the macropixels of an event.
_fill_
Filling factor.

The output ascii file name is the same as the input file name, minus the fits extension of course, and appended with

_smallevents.out
. The Fits output file is appended with
_smallevents_out.fits
. For example if the input file was named
ImageDataCube.fits
, the output files would be named
ImageDataCube_smallevents.out
and
ImageDataCube_smallevents_out.fits
for the ascii and fits outputs respectively.

  • current level of completion:

  • The current algorithm is general in nature, and will accept any 3D data cube, comprising a time series of 2D images. (The images must be of the same wavelength for the algorithm to function correctly. Future work on the algorithm will also include development of the necessary IDL GUI components and their linking to the C-code executable.

Science Test Cases

Case 1: Simulated data, no noise

Description

The first case is actually a series of tests that will use several simulated datasets without statistical noise. Each event is characterised by the following set of parameters: position (xk, yk), spatial size (dxk, dyk), peak intensity Ipk, duration dtk, k=1,…,N. All pixels comprising an event are assumed to be synchronised in time (have the same time profile). A triangular time profile is assumed for all events. The spatial intensity profile is triangular in x and y, where dxk, dyk is the base of the triangle. Background = a non-zero value.
(a) a single event, arbitrary position
(b) N events, arbitrary positions, identical event duration, identical event size, random distribution of event peaks
(c) N events, arbitrary positions, identical event duration, identical peaks, random distribution of spatial size
(d) N events, arbitrary positions, identical spatial size, identical event peaks, random distribution of event durations
(e) N events, random distribution of all parameters. All parameters are statistically independent.

Input

Several input test files were produced for varying image sizes, which included single and multiple events (2,3,5,10 and multiples of 10 events). The image sizes ranged from 8x8, 16x16, 32x32, 64x64,128x128 and 256x256 pixels. Arbitrary constant values for the background were used.

Output

Two output files were produced in each case - An ascii file and fits file. The fits file contained a copy of the original image and an ASCII TABLE extension containing positions and various parameters detailed above. In each case the correct event details were obtained as compared to the input values used to produce the test data.

Case 2: Simulated data with background and statistical noise

Description

The second series of tests will use simulated datasets with the addition of background and statistical noise. Events are defined as in Case 1.
(a) N events, arbitrary positions, identical event duration, identical event size, random distribution of event peaks. Noise added to intensities in all pixels.
(b) N events, arbitrary positions, identical event duration, identical peaks, random distribution of spatial size. Noise added to intensities in all pixels.
(c) N events, arbitrary positions, identical spatial size, identical event peaks, random distribution of event durations. Noise added to intensities in all pixels.
(d) N events, random distribution of all parameters. All parameters are statistically independent. Noise added to intensities in all pixels.

Input

As above. Background pixel values were randomly chosen within a narrow range, such that the lightcurve peak value was greter than or equal to the 3sigma value of the background.

Expected Output

As above.

Case 3: TRACE data

Description

The event detection algorithm has been originally developed to analyse TRACE data. The TRACE dataset used by Markus Aschwanden is re-analysed to verify that it can cope with real instrument data and reproduce Markus’ results.

Input

  • qs171ad.fits contains a TRACE data cube at 171 Angstroms.
  • size of macropixel (length of one side of macropixel)
  • FOV restriction (for beta version, select 'y')

Expected Output

  • qs171ad_smallevents.fits - fits file of small event detection data
The correct Fits and ASCII output files were obtained as expected.

Running the algorithm from AstroGrid

AstroGrid workflow instructions:

  1. Open AstroGrid workbench and click "Task Launcher"
  2. Tasks: find application, specify variables and file as input, specify files as output, launch
  3. Task Launcher search: small events or "Solar Small Event Detection"

-- ElizabethAuden - 04 Aug 2006

Edit | Attach | Watch | Print version | History: r10 < r9 < r8 < r7 < r6 | Backlinks | Raw View | More topic actions
Topic revision: r10 - 2007-12-21 - ElizabethAuden
 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback