Talk:RHESSI Simulations of Complicated Flares

From RHESSI Wiki

What were you people drinking when you made the Clean image for this flare? Here's what I got with Clean and pixon using the RHESSI X-ray observations for this flare.

Of course, I did "cheat" by convolving the Clean components with a Gaussian only half as wide as the Clean-beam, and I used all nine detectors instead of 1, 3, 4, 5, and 6 that you used. I don't understand why you wouldn't use all detectors. Certainly detector #2 provides useful information on the finest structure since this image is for the 25 - 50 keV energy band. In any case, it seems to me that the Clean and pixon images agree very closely so I don't think Clean has met its match at all. In fact, I have not yet seen a flare where CLEAN and pixon images differ significantly as far as the compact sources are concerned.

Given this excellent agreement between pixon and clean X-ray images for compact sources, I am perplexed as to why you would get such a very different Clean image with your simulated data taken from the TRACE white-light image.

Brian Dennis NASA GSFC

CLEAN/Pixons

Hi Brian

thanks for your comments on this. I too was surprised at what came out of CLEAN so we did rerun to check the results. Will do so again as it is completely possible that we messed up. The nice thing about the wiki is that we can edit out the mistakes, I guess.

As for the restricted set of grids - when we did the original HXR imaging work for the paper back in 05 or so the general word on the street was that one should not use grid 2. Certainly not for spectroscopy, and it seemed to be discouraged for imaging as well. And pixons just took so long to run (we had a set of 9 flares to do at different times/different energies) that not using these coarser grids, which wouldn't tell us much about the fine scales we were interested in, seemed a reasonable compromise. Then to be consistent with the HXR work, we stuck to the same grid choice for the model flare too.

best wishes, Lyndsay

What about the halo?

Great! I love to see discussion carried on at such an advanced level.

I'll let Lyndsay provide the parameters of the CLEAN image in the Nugget, but if I did it the only difference between it and the pixon image was to change "pixon" to "clean" in Jim's script.

The CLEAN image that Brian made clearly differs in substantial detail from the pixon image in the nugget, so that needs to be recognized. Maybe Brian can do a nugget on CLEAN/pixon comparisons that quantifies the differences and reveals to the world his tricky post-CLEAN convolution?

Another interesting aspect of this Nugget was the "halo" that appears close to the bright sources. This halo appears in the simulations and there is something similar in the actual images. I'd like to hear Brian's opinion of this. His figure doesn't identify the contour levels but I note an artifact in the contours that is not present in the image, namely the isolated sources to the N.

Hugh Hudson SSL