2. Observational data
The first observational dataset reported here was obtained with SUMER on-board SOHO on 10 July 1996 for O VI 1032 Å from an active region SW on the solar disk (see Fig. 1). The observational sequence for this line was designed so that after an integration time of 15 s, the slit was stepped 1.1 arc sec eastward, accumulating a arc sec2 image in 15 min (see Table 1). After a preliminary analysis of these data we found a continuous series of raster positions with evidence of explosive events between 630 arc sec and 650 arc sec in the E-W raster direction (see Figs. 7 and 8 later).
Table 1. Description of observational data
The second dataset for C IV 1548 Å was obtained from a region in the northern polar coronal hole (see Fig. 2) on 14 July 1996. For this line the integration time was 20 s and the observational sequence was designed so that after each integration, the slit was again moved 1.1 arc sec eastward, thus rastering a arc sec2 area in 10 min (see Table 1). In the present dataset a continuous series of raster positions was found, after analysis, to show evidence of explosive events over a region of 8 arc sec along the slit (N-S direction) and 8 arc sec in the E-W direction (see Figs. 5 and 6 later). Also Si II 1533 Å (formed below K) was observed in this dataset simultaneously with C IV 1548 Å. These profiles were not affected by any explosive event signature, although its intensity was quite weak in the observed region. As has been shown by previous observations, the incidence of explosive events seems to be limited to transition region lines.
The spatial resolution of SUMER is approximately 1 arc sec in the E-W raster direction and 2 arc sec along the slit, N-S direction. Observations were made in the order of diffraction with the corresponding dispersion of 42-44 mÅ/pixel for the aforementioned wavelengths. The coronal hole observations were taken on the more sensitive central part of detector A, coated with KBr which increases the quantum efficiency by an order of magnitude in the range 900 Å to 1500 Å, while the active region observations were taken on the bare part of the detector. We used the arc sec slit for the 14 July dataset and arc sec slit on 10 July.
2.1. Data reduction and further treatment
For the SUMER instrument, the process of data reduction involves three main steps: flatfielding, destretching and radiometric calibration. Our dataset were automatically flat-field corrected on board SUMER. Destretching of the SUMER dataset is necessary, in particular for the data towards the end of the slit due to various wavelength and spatial distortions in the detector (see Siegmund et al. 1994, Wilhelm et al. 1995). For the above corrections the basic IDL routines can be found from within the SUMER s oftware tree.
In order to improve the fitting of individual spectral profiles it was necessary to apply a filtering process. To preserve our spatial resolution (approximately 2 arc sec N-S and 1 arc sec E-W direction), we analyse individually each position along the slit for each raster position. For the raster positions showing explosive events, the signal-to-noise ratio is good although high frequency noise as still present in the profiles. In order to remove this, an optimal (Wiener) filter was applied to the data (Gray 1992). In Figs. 3 and 4 we show examples of the un-filtered and filtered O VI and C IV profiles for a single pixel during each of the explosive events.
In order to calculate a reference profile, only raster positions free of explosive events were selected, averaging them first along the raster direction (E-W) and afterwards along the slit (N-S). The central wavelength of this reference profile is our rest wavelength, from which we measure velocity shifts. However, the rest wavelength of the observed lines is not constant along the slit. In particular, in the active region dataset the average fluctuations are of the order of 5 km s-1 within an area of 5 arc sec and can reach up to 18 km s-1 difference between the intense and darker regions along the slit area. To remove this variation, all the raster positions not showing evidence of explosive events (28 for O VI and 20 for C IV ) were summed together and a mean wavelength as a function of position along the slit was derived. This difference in wavelength was then applied to the individual spectral profiles thus standardising them all to a common wavelength. The above process was then repeated in order to check the accuracy; an uncertainty of 2 km s-1 (1) was derived.
2.2. Contemporary observations with SOHO
In order to contrast our observations we have checked for contemporary observations done with other instruments on board SOHO, Yohkoh and also ground observations. The most interesting comparison comes out for the data taken on 10 July 1996. From comparison with Yohkoh, EIT and CDS images we can see that our SUMER raster covers an area near the apparent foot points of a complex structure of loops in an active region located SW on the solar disk (see Fig. 1). The structure of this active region remains for more than two days, from 9 July to 11 July 1996.
A Mt. Wilson magnetogram was taken nine hours later than our observations. Our raster coincides with a bipolar magnetic feature in the middle of an active region. At the north of our raster image the emerging magnetic field is weaker than at the south. It is in this northern region where the velocities of the observed explosive events are higher and their incidence prolonged, as will be described in the next section. Other authors have associated explosive events to the cancellation of phostospheric magnetic flux and preferentially regions with weak and bipolar magnetic field (Porter et al. 1987, Dere et al. 1991, Moses et al. 1994, Chae et al. 1998).
© European Southern Observatory (ESO) 1999
Online publication: December 22, 1998