DIY spectroscopy: Analyzing AIA diffraction patterns

From RHESSI Wiki

Revision as of 14:06, 28 May 2011

Introduction

The EUV portion of the spectrum reflects the physics of the solar chromosphere and transition region and in this Nugget we show how an unheralded design feature of the SDO/AIA instrument allows us to study flare spectra pixel by pixel. With the typical approach to EUV spectroscopy - a scanning slit - we sacrifice resolution in time and space for the sake of spectral resolution. With direct EUV imaging, the main purpose of AIA, we can achieve superb cadence and spatial resolution but we sacrifice spectral response. Finally, with an irradiance instrument like EVE, we can obtain the spectral resolution and range along with very high cadence, but at a cost of not having any spatial resolution at all. Here, we are striking a different balance by taking saturated AIA images and doing DIY spectroscopy: we are using the artifacts of diffraction and dispersion to deconvolve the spectrum observed by AIA from its original single-flux-per-pixel value into the original spectral line components.

Figure 1: Diffraction patterns observed in the AIA 193 Å passband. The diffraction spikes, in this case, extend out into the corona right out to the limit of the AIA field of vew (inset). Note that each blob within the pattern is actually an image of the source, dispersed in wavelength.

Technique

Diffraction patterns can frequently be seen in AIA images, such as in Figure 1. These are caused by the front filter grid support mesh behaving as a pair of perpendicular diffraction gratings, forming a pair of orthogonal diffraction patterns. Since each telescope houses two front grids, we observe eight diffraction arms, as in Fig. 1. Under normal circumstances, the shadow of the grids is removed though image pre-processing and the intensity is not sufficient to generate a diffraction pattern. However, in cases where the intensity in a pixel (or group of pixels) becomes very high, diffraction patterns can be observed. Since this is most often observed when the intensity saturates the CCD, understanding the context of the AIA source images is often difficult. In a break from the norm, we use RHESSI data for context imaging. In this way, we can identify where along a loop the diffraction pattern originates and can therefore analyze the diffraction pattern as a function of both time and space (perpendicular to the diffraction angles at least).

In a similar way to a prism dispersing white light into its colour components, dispersion is observed within diffraction patterns. Close to the source, the bright diffraction peaks are narrow but as higher orders of diffraction are observed, the bright peaks become more spread out. This is due to emission at different wavelengths (λ) being diffracted by different amounts. The distance from the source r_m at any order (m) can be translated in to wavelength as λ = (r_m x a)/m given a, the spacing of the diffraction grid, in this case, the wire mesh (Lin et al. 2000). It is clear that the dispersion within a single order, δr and therefore the spectral resolution, δλ, will increase with diffraction order m. Thus, the large field of view provided by AIA offers a significant improvement on the spectral resolution available previously from TRACE diffraction patterns. Fig. 1 shows diffraction out to 52 orders, compared to ~23 orders observed with TRACE (Lin et al. 2000). This means that at the 52nd diffraction order in the 193 Å passband, 1 Å spectral resolution corresponds to 3”, or 5 pixels.

Figure 2: Top: AIA 193 Å dispersion spectrum (black) with background-subtracted EVE spectrum. Bottom: AIA 193 Å dispersion spectrum (black) with two isothermal CHIANTI spectra synthesized at 107 K (red) and 106.2 K (green) along with their combined contribution (dashed).

With a spectral resolution of up to 0.7 Å, we can resolve individual spectral lines within the dispersion spectrum. To help identify the lines of interest, we can compare the dispersion pattern to SDO/EVE spectra (Fig. 2, top). This is especially helpful in identifying blended lines, which we may not otherwise resolve. With the primary lines of interest identified, the dispersion spectrum can be modeled using a combination of isothermal CHIANTI spectra, such as those shown in the bottom panel of Fig. 2. The scaling factors (1.8x10⁵⁰ and 6x10⁴⁷ cm^-3 for the high and low temperature component respectively) correspond to the emission measure of the plasma at that temperature. Repeating this process for many passbands will give us access to the temperature and emission measure of a range of spectral lines which, when combined, will result in the differential emission measure (DEM) of the plasma in saturated flaring pixel.

Conclusions

Applying this technique to all diffracted AIA images at the high cadence of 12 seconds we can investigate how the DEM function varies with time at various points within the flare. While the spectral resolution and range may not compare to the likes of EIS or EVE, we are still achieving <1 Å resolution. This is a unique way of optimizing the compromises required to investigate the dynamic nature of a flare DEM function. While this analysis is still in the very early stages, it will provide very useful and detailed results in the future.