Weekly Report 23Jul2010 30Jul2010

From RHESSI Wiki

Jump to: navigation, search

Contents

Weekly Report 23Jul2010 30Jul2010

Simulated Point Source with Albedo Imaged Using Forward Fit Techniques

An eventlist for a point source with albedo was created for an event near limb. As shown in the previous report various imaging techniques were used to reconstruct the RHESSI images. This report focuses on using Forward Fit Techniques to image the albedo and it's foreshortening near the limb.

VIS FWDFIT with two ellipses

Using the VIS FWDFIT Image reconstruction algorithm a point source and albedo near the limb were reconstructed using a fit with two ellipses. The background image is the data map that was used to construct the eventlist, the blue contours are the vis_fwdfit image.

The background map is the data used to create the eventlist. The blue contours are the vis_fwdfit reconstruction with two ellipses as the initial guess at the shape of the image.

The specifications for the image are:

The image was created using information about the approximate location of the point source and the albedo along with their sizes from the original data map as the initial guesses for the two ellipse fit. The model of the point source is off center and larger than the original source. The albedo is offset and from the original and extended along the Heliocentric Y Axis. The eccentricity of the ellipse modeling the albedo is extremely high. Attempts to create this image using different detector combination's and different fluxes and locations of the initial guesses for the ellipses caused the algorithm to fail. Removing detectors 1 and 2 caused a complete breakdown in the ability of the algorithm to image the albedo component. In those cases the point source could still be imaged with varying degrees of accuracy but the albedo could not be imaged at all.

The profile along a line parallel to the Heliocentric X Axis. The red trace is the original data map profile. The green trace shows the profile of the reconstructed image using vis_fwdfit with two ellipses as an initial model.

The profiles along a line parallel to the heliocentric X are shown above. The point source in the Reconstructed image is offset from the data map by .5 to 1 arcsec. The point source is also larger in the Heiolcentric X direction. The albedo contribution is offset by approximately 1 arcsec. The peak flux of the contribution is also larger in the model. These artifacts appear when trying to use two ellipses to force the reconstructed image to model the albedo. When using the vis_fwdfit image reconstruction technique with a single ellipse the point source is modeled as a single compact source and the extended albedo is ignored. This implies that the using vis_fwdfit with two ellipses is not a good model if the sources are close together and one is weaker and more extended than the other.


VIS FWDFIT with a single ellipse - Comparison of visibilities at different angles

The reconstructed images below shows a vis_fwdfit reconstruction with a single ellipse as the initial guess. The difference between the two image reconstruction is the guess of the angle of the minor axis of the ellipse with respect to the heliocentric x axis. Note that for a guess with an angle of 90 degrees if the fit algorithm is allowed to go through enough iterations it will rotate the image so it agrees with the algorithm using an initial angle of zero if all other initial parameters are the same. In the image below the green contours show the effect of using an angle of 90 degrees with the parameters set to final values for a fit with an angle of zero and no iterations done.

The statistics and parameters for the fit below are:


The background image is the original data map of a point source with albedo contribution near the limb. The blue contours are a single ellipse reconstructed image using vis_fwdfit with an angle of zero. The angle measures the angle of the minor axis with respect to heliocentric x. The green contours are the same vis_fwdfit reconstruction with an angle of 90 degrees.

The reason for doing this is to look at the fits to the visibilities that are used when reconstructing the image. Before that is considered they are a few points mentioned in the two ellipse fit that should be addressed here. Consider the blue contours of the single ellipse fit compared to the two ellipse fit in the first figure. With a single ellipse fit the image is centered on the original point source. The angle of the reconstructed image when using the single ellipse fit is much closer to the angle of the actual data map (~0 degrees)than in the two ellipse case. The single ellipse image does not attempt to image the shape of the albedo but gives a much more accurate image of the brighter point source. Using this data set it seems that vis_fwdfit works best to fit a single a bright source but is not effective for imaging any nearby weak extended sources. Since the albedo would appear as a weak extended source it may be beyond the scope of vis_fwdfit to image the albedo.

The reconstructed images at different angles shown above were created to test the fits to the visibilities. As the angle is changed the peaks and valleys of the modulations in the fit to the the visibilities should shift. Below are shown the plot of the light curves and fits along with C statistics for each fit as well as the fit to the visibilities.

The profiles for each detector along with the fit and the C statistic values for the vis_fwdfit image with an angle of 0.
The visibility data for the simulated point source with albedo near the limb along with the fit to the visibilities for a fit with an angle of 0 degrees
The profiles for each detector along with a 'fit' and the C stastics for a vis_fwdfit image with an angle of 90 degrees and all other paramaters equal to those for the fit with an angle of 0 degrees
The visibility data for the simulated point source with albedo near the limb along with the fit to the visibilities for a 'fit' with angle of 90 degrees and all other parameters fixed at the values for a fit with an angle of 0 degrees.

First consider the profiles for each detector in the two cases. The C stastics are close to 1 for all detectors in both cases. The difference in the overall c statistic is in the second decimal place, hence it is not statistically relevant. Now looking at the plot of the visibilities note that the measured visibilities are the same, however the fits in the finer detectors are not. For the coarser detectors (detectors 4-9) the modulation information is lost because of size of the source. However for the finer detectors the shift in the visibilities is apparent. The peaks at an angle of 0 degrees become the valleys at an angle of 90 degrees and vice versa. Changing the phase of the visibility measurements may not show up in the C statistic but it is important in the creating the image as the contours of the two images show.


FWDFIT using a elliptical sources sources

The following image was created using the fwdfit algorithm. The specifications for the images are:

The background is the original data map. The blue contours show the fit using a single gaussian ellipse, the green contours show the fit using two gaussian ellipses.

The background image shows the data map of a point source with albedo near the limb. The blue contours show a reconstructed image using FWDFIT with a single elliptical source. The green contours show a reconstructed image using FWDFIT with two elliptical sources. All other parameters are the same for the two images

The image below shows the observed counts and fits for each detector along with the C statistics for a single gaussian ellipse fit.

The individual light curves for each detector are shown along with the fit using FWDFIT with a single elliptical gaussian source. The C statistics for each detector are shown.

The image below shows the observed counts and fits with the C statistics for a two gaussian ellipse fit.

The individual light curves for each detector are shown along with the fit using FWDFIT with two elliptical gaussian sources. The C statistics for each detector are shown.

In both cases the C statistics are good. Looking at the contours the the single gaussian ellipse is more compact and centered on the source. Once again it does a good job of imaging the bright compact source but does not image the faint extended albedo source. The double elliptical gaussian looks to be a better fit to both sources. It is still centered around the bright compact source but its shape is not symmetric along any axis through the bright source. The image is deformed as can be seen by the contours crossing each other on the right of the image. If the extended albedo source was brighter than in this case there is the possibility that this method could image it.

To test this we need to create an eventlist for a source with an artificially high albedo contribution and try to image that.

Moving RHESSI Web Content

Work is under way on moving RHESSI Web Content to the new Content Management System, about 1/4 of the top level pages have been move (top level being one link from the home page) however because of the hesperia security violation last week, we were unable to might our goal of having the new site ready for August 1st. We should be back online this week with a new target date of August 12th for completion of migration.

Goals for this week

Personal tools
Namespaces
Variants
Actions
Navigation
Toolbox