The fortunate position of the Galileo spacecraft provided us with a un
ique opportunity to directly observe the Shoemaker-Levy 9 impacts as t
hey occurred on the far side of Jupiter, and we present observations o
f the G fireball obtained by the Near Infrared Mapping Spectrometer (N
IMS). These measurements were performed using 10 spectral bands, 4 rep
resenting continua and spanning the wavelength range 1.84 to 4.38 mu.
Fireball signals were evident for up to 80 sec, with the time of inten
sity maxima and duration being greater for longer wavelengths. Color t
emperatures and effective emitting areas were estimated by fitting bla
ckbody functions at the four continuum wavelengths. Good blackbody fit
s were found, and their intensities at shorter wavelengths show excell
ent agreement with the Galileo Photopolarimeter/Radiometer measurement
s. Temperatures near the beginning are above 3000 K, decreasing to sim
ilar to 1000 K after 1 min. The corresponding areas range from 400 to
20,000 km(2). The effective diameter of the luminous fireball shows ap
proximately linear time variation, at least for the first 45 sec. From
the temperature-effective diameter relation we find an adiabatic coef
ficient of gamma = 1.2 +/- 0.1, much as expected from theoretical cons
iderations. The luminosity, when integrated over the period of observa
tions and assuming a Stephan-Boltzmann radiator, gives an above-cloud
radiative energy loss of 0.48 +/- 0.13 x 10(25) erg. As a conceptual a
id, we developed a simple, heuristic theory of the fireball phenomenon
, considering the penetrating fragment's wake (termed debris channel)
to consist of high-temperature jovian and cometary material, which und
ergoes radial expansion and acceleration back along the wake axis. The
outer layer of the material in this debris channel is presumed optica
lly thick, radiating as a blackbody to produce the observed emissions
(we speculate that the opacity is produced by condensed refractories s
uch as MgO and SiO2, probably containing impurities). One-dimensional,
variable-area axial flow of a radiating, compressible, inviscid gas i
s concurrently solved with the radial shock motion occurring in the no
n-axisymmetric atmosphere. We calculate debris surface radii and veloc
ities using Sedov's theory for line explosions. The assumed initial de
bris surface temperature is consistent with entry shock heating. Our s
imple model shows good agreement with the observations, for both the t
emperature and luminous area, and suggests that the diameter of the G
fragment (assumed spherical and of unit density) was 300 +/- 100 m, wi
th a nominal energy of 2.5 x 10(26) erg. The measured luminous energy
is within a factor of 2 of that predicted for the nominal impactor siz
e, whereas the amount of water in the splashback, as measured by G. L.
Bjoraker et al. (1996, Icarus 121, 411-421) and Th. Encrenaz et al.,
(1997, Planet. Space Sci.) agrees to a factor of 3 with the model resu
lts; however, the large CO abundance obtained by E. Lellouch et al. (1
995, Nature 373, 592-595) is inconsistent with the suggested size. Thi
s diameter estimate must be considered provisional and probably a lowe
r limit; precise estimates require comparison of the measurements with
comprehensive numerical simulations, which we encourage. (C) 1997 Aca
demic Press.