E-mail: aschwanden@lmsal.com - Markus J.Aschwanden (Lockheed Martin Solar & Astrophysics Lab.)

RHESSI Flare Observations: 2002-Feb-20, 11:06 UT

Since the launch of RHESSI on 2002-Feb-5 we have real solar data. Some 20 solar flare events have been observed during the first two months. One of the most prominent ones is the 2002-Feb-20, 11:06 UT, flare (GOES C7-class, AR 9825), which was featured in the first NASA/RHESSI press release on March 20, (along with material from a second flare at 21 UT in AR 9830).

NASA Press Release

Brian Dennis provided the following URL sources:
You can view yesterday's NASA Press Release with the RHESSI first-light movie at the following location:
http://www.gsfc.nasa.gov/topstory/20020320hessixray.html
First images, including another great movie, that were part of the Swiss press release are at
http://www.hessi.ethz.ch/pop/Bilder/FirstImages.htm
The press briefing material for the launch is still online at
http://www.gsfc.nasa.gov/gsfc/spacesci/hessi/animations.htm
You should also check out the new RHESSI Data and Software Center at
http://hesperia.gsfc.nasa.gov/hessidatacenter/index.html
You can reach this from the Goddard RHESSI home page -
http://hesperia.gsfc.nasa.gov/hessi/
by clicking on the Data/Software link.

BBSO and EIT context images

Additional images of the flare location of the 2002-Feb-20, 100:06 UT, flare can be seen from Peter Gallagher's BBSO MaxMillenium webpage
http://www.bbso.njit.edu/arm/20020220/9825.html
according to which the flare took place in Active region AR 9825, at the heliographic location N16 W80 (close to the limb in the NW), listed as C7.5 flare (at 11.07 UT)

WIND observations

Context information of WIND observations simultaenously obtained with RHESSI are shown on Sam Krucker's webpage
http://sprg.ssl.berkeley.edu/~krucker/hessi_plots/

RHESSI imaging with CLEAN

A tutorial on initial imaging of this flare data with the CLEAN algorithm is also provided on Sam Krucker's webpage
http://sprg.ssl.berkeley.edu/~krucker/hessi/clean_real.html
containing information on

RHESSI imaging with MEMSATO and MEMVIS

Images of this 2002-Feb-02, 11:06 UT, flare have also been produced by Andrew Conway, using the MEM-VIS and MEM-SATO algorithm, in the 12-25 keV energy range, with the specifications These high-quality images reveal not only the dominant double-footpoint sources, but also the thermal emission in the overlying flare loop. The two images, communicated by Andrew Conway on April 10, are shown here:




RHESSI imaging with Forward-Fitting

Here we describe the first attempts of imaging with the Forward-Fitting method. We select an interval around the flare peak time, 2002-Feb-20, 11:06:00 UT, with a duration of 40 s or about 10 rotation periods. The RHESSI light curve from the entire energy range of 3-15,000 keV is shown in the following middle plot, with the selected 40-s time interval indicated. For comparison, we show also the GOES light curves (Lo, Hi channel), and their time derivatives (dashed), which should mimic the hard X-ray light curve if the Neupert effect applies (top panel). In the bottom panel the variation of the spin rate is shown, as determined with PMTRAS, having an average of r=4.332946 s in the considered time interval (bottom panel).


An example of a command-line run for forward-fitting a RHESSI image is:


We repeat imaging with forward-fitting in the same time interval at different energies, i.e. 10-12 keV, 12-14 keV, ..., 26-30 keV, 30-40 keV, ... , 70-80 keV, covering the entire energy range from 10 to 80 keV. A combined plot of all forward-fitting images in these 14 energy ranges are shown here:


The first 10 energy intervals are identical with the CLEAN images shown in the press release (URL given above).


Next we produce a time sequence of images, starting at flare begin around 11:04:41 UT, stepping in intervals of 15 s until 11:07:56 UT, always in the energy interval of 18-80 keV. The sequence shows that bright double footpoint sources are only seen during the interval of 11:05:56-11:06:26 UT. The reconstruction in the prior and later time intervals does not always reveal double footpoint sources as expected, but rather diffuse larger source at times. This does not necessarily mean that double footpoint sources do not exist before or after the flare peak, the lack of their reconstruction could be an artifact of insufficient count rates, when information on spatial scales gets lost.


All previous RHESSI images do not show the flare at the correct heliographic position, because the RAS solution was not yet provided in the software at this time. However, the BBSO and EIT images indicate the true flare location at position [919", 285"] west and north from Sun center. This implies that the RHESSI images have to be rotated by a angle of 208.1 deg (clockwise), and an additional offset of [16.6",5.2"] exists between the RHESSI and BBSO flare position (which could perhaps be due to a different definition of flare positions). The RHESSI flare position is measured here at the midpoint between the HXR high energy footpoints. A map of the rotated flare position is shown below. The lower panel shows an overlay of forward-fitting maps with contours at half maximum in the 14 energy channels between 10 and 80 keV. The centroids of the forward-fitted gaussians are marked with crosses.


One of the foremost strength of the forward-fitting method is that the spatial positions of the centroids of the fitted gaussian sources have the highest possible absolute and relative accuracy, when compared in images with different energies or time intervals, if the same model is fitted. The centroid position of local flux peaks in images reconstructed with other algorithms is generally less robust because global image solutions are also sensitive to the overall flux distribution in the image, which can contain sidelobes and noisy background. The present flare occurred very close to the limb, so that a radial shift in the source position (from Sun center) translates directly into an altitude difference. In the following image we show the centroids of the footpoint sources as fitted with gaussians, indicated with diamonds for the southern footpoint, and with crosses in the northern footpoint, for the 14 images with energies between 10 and 80 keV. We see that there is a systematic altitude shift with energy. The bottom panel shows the altitude difference as function of energy, amounting up to 4000 km between 20 and 80 keV. The solid and dashed lines are subject to 3-point smoothing. The fact that they show a monotonic trend of decreasing altitude with higher energies speaks for a systematic real trend, measured with an accuracy of about 0.5". For flares with higher photon count rates we expect even higher accuracies.


This example illustrates a useful application of forward-fitting, namely the measurement of the height dependence of energy loss in flare loops as function of energy. Such measurements have only be performed with very crude accuracy before, due to the lack of energy resolution and insufficient photon statistics. A statistical study of relative flare H-alpha positions with respect to hard X-ray flare positions revealed height differences of 9700 km, 8700, and 7700 km between the 4 Yohkoh/HXT energy channels (14, 23, 33, 53 keV), as measured by Matsushita, Masuda, Kosugi, Inda, & Yaji (1992, PASJ 44, L89). For the limb flare of 1993-Feb 17, 1035:44 UT flare, absolute heights of 5900 km, 4000 km, and 3500 km have been determined for the three lowest Yohkoh/HXT channels (Aschwanden, Fletcher, Sakao, Kosugi, and Hudson 1999, ApJ 517, 977). Theoretical modeling for the height distribution of nonthermal hard X-ray sources in impulsive solar flares is described in Fletcher (1996, AA 310, 661).

E-mail: aschwanden@lmsal.com - Markus J.Aschwanden (Lockheed Martin Solar & Astrophysics Lab.)