Clean and MEM_NJIT Compared

These figures show that the MEM_NJIT "super resolution" is about equivalent to using one finer RHESSI subcollimator. The first figure shows three double-component flares (rows 1, 2, and 3) mapped using the Clean algorithm with grids 2-8, 3-8, and 4-8 in columns 1, 2, and 3 (respectively). The instrumental resolutions of the 3 columns are, respectively, 4.3, 6.9, and 13 arcsec.

FIG 1. Clean maps of 3 flares with 3 different sets of subcollimators (D 2-8, D 3-8 and D 4-8).
Contours start at 10% of maximum, increasing at 10% increments to 100%.

The second figure shows the same flares (rows 1, 2, and 3) mapped using MEM_NJIT with uv circles 2-8, 3-8, and 4-8 in columns 1, 2, and 3 (respectively). Compare the middle MEM_NJIT map in the first row of Fig. 2 with the 2nd Clean map in the first row of Fig. 1. The two maps are similar.

Comparison of the middle MEM maps with the column-1 Clean maps again reveals similarities. (The Clean maps show some low-level (10-20%) noise that is probably due to extended structure that is resolved out ("over resolved") by the finest subcollimator. MEM_NJIT appears less susceptible to this.)

Comparison of column-3 MEM maps with column-2 Clean maps does not show the same degree of similarity found above. The column-3 MEM maps of the first two rows show only one component. Thus "super resolution fails in these cases. The 3rd flare mapped by MEM_NJIT, however, shows two components much like the Clean map (column 2) with higher angular resolution. The relative fluxes of the two sources are reversed between Clean and MEM_NJIT: in the former, the southern scomponent is brighter, while in the latter, the northern component is brighter.

FIG 2. MEM_NJIT maps of 3 flares with the equivalent uvcircle sets (UV 2-8, UV 3-8, and UV 4-8) used in Clean

CONCLUSIONS

We conclude that the MEM_NJIT "super resolution" more clearly separates components that are not well separated by Clean. This result is similar to that found by Cornwell and Evans (1985) where they compared VLA maps made using MEM in the C configuration with maps made in the B+C configuration. In their case, MEM maps were similar to the higher resolution Clean maps. However, we find that the flux ratios of the components are not faithfully determined by MEM_NJIT when the instrumental resolution is not adequate.
Last modified: Tue Nov 15 13:58:17 EST 2005