4. Discussion and comparison with near-infrared lightcurves
We can estimate the atmospheric depth where the emission arose at the last maximum in the CCD lightcurves. For L, the flux at 892 nm was 1.05 0.15 10-11 W/m2 /µm (calibrating the lightcurve presented by Schleicher et al., 1994 in the way described in the note on absolute calibrations and adding a 10% uncertainty associated with this procedure) and 1.22 0.09 10-11 W/m2 /µm for 907 nm (Fitzsimmons et al., 1996).
We can provide an estimation of the methane abundance by the following means: Using (µ=0.15 0.02), derived from the expected position of the ballistic ejecta as a function of time, and the methane absorption coefficient for the appropriate wavelength (), with computed for the spectral resolution of each measurement by convolving methane absorption coefficients (Karkoschka, 1994) with their respective filter transmission curves, we can determine the abundance (a):
assuming . I is the observed intensity and is a given time. Solving for a we get a =2 m-Am of methane, which corresponds approximately to 1 km-Am of jovian atmosphere, assuming a constant methane molar fraction of 2 10-3. This corresponds approximately to the 25-mbar pressure level. We have computed the fractional error in a due to the propagation of errors in intensities, absorption coefficients and µ. The resulting 1-sigma fractional error is 2. This large error arises mainly from the uncertainty in the absolute fluxes. The resulting uncertainty in pressure gives 25 mbar.
For H, the flux at the 892-nm filter shown in Fig. 1b is (4 1) times lower than that at 948 nm, which would imply thermal emission from the 60-mbar level. On the other hand, there is a 2-min time difference between the analyzed fluxes, which is probably quite significant. The time difference would reduce by a factor of 2 or 3 the 948-nm absolute flux at the time of the 892-nm observation, if we assume a lightcurve similar to that of Schleicher et al. Assuming a decrease by a factor of 2.5, we get an abundance of 0.003 km-am of methane, corresponding to an emission level located at 35 mbar (note that the filters used in the H observations were much narrower than for L). By estimating the propagation of 1-sigma errors, we derive P= 35 mbar.
One must note that 1 to 20 mbar is the preferred range of pressures for the impact-generated hazes at the outlying regions (West et al. 1996) as determined from post-impact HST observations and is also in agreement with the reflectivities observed at the opaque 2.3-µm region. If the aerosols were generated before their descent, then they came to a complete stop not much deeper than 20 mbar. This is close to the levels that our observations suggest. If this is really the case, then the dark aerosols "survived" the heating of the fall-back without vaporizing. This may give an indication of the kind of composition of the haze aerosols, since many of the proposed solids would vaporize at the fall-back temperatures.
From the fluxes reported for L impact at 907 nm by Fitzsimmons et al., we cannot make an estimate of the temperature of the emitting region unless we assume a certain solid angle for emission or have another flux value at another wavelength in the visible. To be detectable in the visible, the second maximum would require temperatures in excess of 1800 K, assuming Stefan-Boltzman law and very good observing conditions (with observing conditions as favorable as those described here the flux at 550 nm should be larger than 10-12 W/m2 /µm). In the near future we plan to conduct a search for detection of this second maximum in several data sets as well as to model the outgoing radiation from an emitting-absorbing-scattering medium.
The order of magnitude difference in flux emission for L and H may be related to very different temperatures of the plumes or to the different area of emission, or both. Our preferred explanation is that the difference is a result of the different amount of material ejected in the 9-12 km/s velocity range.
The steep rise in the 2.3-micron near-infrared curves for impacts H and L, 49 12 sec and 39 3 sec after impact respectively (the begining of the second precursors) can be explained by emission from the highest velocity ejecta reaching the level of the limb as seen from the Earth. These ejecta must have v=9 km/s (see Fig. 5). Velocities higher than this would have resulted in earlier detections. The impact times were adopted from the Galileo PPR lightcurve (Martin et al., 1995) and the near-infrared lightcurves and timings are those by Hamilton et al. (1995). From this determination, the fastest material which is optically thick enough to radiate measurable quantities is moving at 9 km/s, the same as the determination made using the first detection of the plume in visible light (9 1 km/s).
The maximum in the second precursor may occur when no additional material can reach the limb level (ejecta slower than 3.5 km/s). In Fig. 5 we see that no further material can reach this level after 200 s. Then, the maximum should be around this time, for the second precursors. However, the plume is cooling while going up, and that may change the predicted behavior by a large amount. We see the second precursor peak about 180 s after H impact which is considerably close to the estimate. The maximum of the L second precursor seems to occur before the predicted time. The begining of the main near-infrared peak would correspond to the fall to the original levels of ejecta with 4 km/s initial velocity.
As we have already pointed out, the 2.3-µm maximum occurs earlier than the CCD second maximum. The 2.3-µm peak may be due to the reimpacting material that was ejected at velocities lower than 9 km/s, which may represent a small fraction of the total plume or may not have enough energy to trigger the visible emission. But the total energy released in the near infrared wavelengths is higher than that at visible wavelengths, which seems to be contradictory with the idea that the ejecta with velocities lower than 9 km/s constitute a small fraction of the plume.
One solution is to hypothesize that the particulates are optically thick in the visible, but the gas is optically thin. As the particulates are optically thick to reflected sunlight, they should also be efficient thermal emitters. If the particulates were the only material which emitted in the visible, the emission could be delayed in this wavelength range compared to the near infrared because the particulates were concentrated in the ejecta with v=11 km/s. It is still a matter of speculation why the particulate ejecta were concentrated in a certain range of velocities, but that supposition is also supported by the arc shapes of the scars, as opposed to a continuous semidisk shape.
An alternative explanation for the difference in the timings of the visible vs near-infrared emission maxima is that the infrared emission took place hundreds of kilometers above the visible emission level. There, the friction might have been enough to radiate in the near infrared, but not in the visible, as the material was not hot enough or was not sufficiently optically thick in the visible.
© European Southern Observatory (ESO) 1997
Online publication: May 26, 1998