We have applied computer stereophotogrammetry to Apollo Lunar Surface Close
up Camera (ALSCC) pictures of the lunar surface to construct the first-ever
digital topographic relief maps of undisturbed lunar soil over spatial sca
les from 85 mu m to 8.5 cm. Using elevation histograms, fractal analysis, a
nd Hapke's photometric roughness model we show that Apollo 14 (Fra Mauro) I
mbrium ejecta is rougher than average Apollo 11 (Mare Tranquilitatis) and A
pollo 12 (Oceanus Procellarum) mare surfaces at submillimeter to decimeter
size-scales. We confirm the early result of K. Lumme et al. (1985, Earth Mo
on Planets 33, 19-29) that the cumulative distribution of elevations for lu
nar soil is typically well described by Gaussian statistics. However, cumul
ative distributions are insensitive to asymmetries in the shapes of elevati
on histograms: Of 11 discrete elevation histograms we measured, about half
exhibit significant deviations from Gaussian behavior. We also confirm Lumm
e et al.'s finding that the roughnesses of all lunar surfaces increase with
decreasing size-scale. We further show that the scale dependence of roughn
ess is well represented by fractal statistics. The rates of change of rough
ness with size scale, represented by fractal dimension D, are remarkably si
milar among terrians. After correcting for the contribution of large-scale
roughness, our average value of D = 2.31 +/- 0.06 falls within the range 2.
0 less than or equal to D less than or equal to 2.4 reported from lunar rad
ar studies. The amplitude of roughness, which we characterize with the rms
slope angle at l-mm scale, varies significantly among terrains. For lunar m
are, the average rms slope angle is 16 degrees +/- 43 degrees and that for
Fra Mauro regolith is 25 degrees rt 1 degrees. By comparison to radar data,
we suggest that the roughness of Fra Mauro (Imbrium ejecta) regolith is si
milar to that of lunar highland terrains. We find that the Gaussian slope d
istribution assumed in B. W Hapke's model (1984, Icarus 59, 41-59) adequate
ly describes typical lunar regolith surfaces. A revised form of Hapke's equ
ation that models realistic particle phase functions and the coherent backs
catter opposition effect was fitted to disk-resolved lunar photometric obse
rvations and yields estimates of <(theta)over bar> = 27 +/- 1 degrees for h
ighland and a = 24 =/- 1 degrees for mare regolith. These values of <(theta
)over bar> as well as the implied relative highland:mare photometric roughn
ess ratio are best matched in our elevation data by the cummulative contrib
utions of surface topography covering all scales greater than 0.1 mm. Less
than 5% of the photometrically detected roughness of lunar regolith is cont
ributed by surface relief at scales larger than 8 cm. This conclusion impli
es that values of 8 derived from whole-disk and disk-resolved photometry, r
espectively, may be taken to represent the same physical quantity. In addit
ion, particulate samples used in goniophotometric measurements should not b
e assumed to be photometrically smooth (i.e., 8 = 0 degrees), as is often d
one in laboratory applications of Hapke's photometric model. The predicted
photometric roughness at size scales of 0.1 mm and less significantly excee
d photometric estimates and suggests that there exists a measurable size sc
ale below which topographic relief either is not photometrically detectable
or is not represented in the Hapke model as macroscopic roughness. (C) 199
9 Academic Press.