2. Observational results
In Figs. 1a, 1b we present the H lightcurve (impact flux versus time) in absolute units at 948 nm as observed from the 4.2-m WHT. We also present a flux value at 892 nm after the second maximum. In Fig. 2, we present the sequence of images which show the plume development analyzed here. The detection of the H impact plume was already presented by Larson et al. (1994).
The first image of the sequence is the first image obtained by the telescope. No previous images were taken because it was twilight. The instrumental set up, used previously (Moreno et al., 1995) includes a 1280 1180 coated EEV P88300 CCD and interference filters centered at 948 nm and 892 nm with a full width at half maximum of 50 Å. Plots of their transmission curves can be found in Ortiz (1994). The plate scale was 0.102 arcsec/pixel. Integration times are shown in the caption of Fig. 2.
To generate the lightcurve in Fig. 1a, we first removed the bias and flatfielded the images. After that we interpolated over bad pixels, fit an ellipse to the limb of the planet, centered, and rotated the jovian images. We selected the location of the plume in each image by visual inspection. This location moved a few pixels mainly along the E-W direction because of both the rotation of Jupiter and the ballistic motion. We integrated the signal within a 0.51 0.51 arcsec square aperture, subtracted the sky contribution and subtracted Jupiter's contribution. The result was divided by the sum of all the data numbers from Jupiter and calibrated in flux as described in the section "Notes on absolute calibrations". The sky contribution was computed using the mean count rate from five aperture positions in the sky. The jovian contribution was determined by placing three apertures 2 arcsec apart from the plume position, sampling only the limb of Jupiter (see Fig. 3).
Sky noise is the main source of uncertainty in our lightcurve. This is shown in Fig. 1 as the standard deviation of the 5 sky positions. Another source of uncertainty was the size of the aperture and the jovian contribution.
The size of the aperture was too small to include all the flux from the plume, which is spread by the seeing. However, a larger aperture size resulted in a considerable increase in uncertainty arising from the sky noise. As seeing remained very stable during the sequence, we estimated that a constant percentage of the flux (20 to 40%) was outside the aperture in all the images. The uncertainty in the removal of the jovian contribution was estimated to be 2 10-13 W m-2 µm-1, derived by computing the dispersion from 3 different apertures located 1 to 2 arcsec from the plume at slightly different positions (see Fig. 3). This uncertainty is more than an order of magnitude smaller at 892 nm. The uncertainty in time is assumed to be less than 1 s because the computer was linked to a GPS receiver. The gaps between observations are the result of the long times needed to read the CCD and the time needed to operate the computer. Between the second and third images analyzed here, two frames were taken by reading only a small "subwindow" of the CCD in order to save time. Unfortunately, this mode did not work properly and the resulting images were not useable.
We obtained a better estimate of the flux by using deconvolved images. The improvement in spatial resolution allows for the use of small apertures. We deconvolved our images using a Lucy (1984) algorithm and a model for the point spread function which worked well for planetary images (Ortiz, 1994). The lightcurve using a 0.51 0.51 arcsec square aperture is shown in Fig. 1b. These fluxes are listed in Table 1.
Table 1. H impact fluxes
Although the time resolution is not very good, the data show the typical pattern already discussed. In the lightcurve (Fig. 1a, 1b) there is a maximum followed by a minimum, and another maximum. The maximum flux at 2.3 µm was observed at 19:44:30 45s UT, (Hamilton et al., 1995), which is 19.74 UT, earlier than the second maximum at 948 nm. The shape of the WHT lightcurve corresponds closely with the maximum, minimum and second maximum in CCD lightcurves for L (Schleicher et al., 1994; Fitzsimmons et al., 1996), also plotted in Fig. 1c. Again, the second maximum in these CCD lightcurves (which ocurred at 22:31:52 according to Schleicher et al., 1994) is observed after the main peak at IR wavelengths, which took place at about 22:30 UT (Lagage et al., 1995). The difference is close to 2 minutes. We note that there is an order of magnitude difference between the peak fluxes of the H and L impact CCD lightcurves.
© European Southern Observatory (ESO) 1997
Online publication: May 26, 1998