Magellan radar scattering from the surface of Venus.

摘要

Magellan spacecraft probed the surface of Venus with 12.6 cm microwave radar over a period of three years. A discretized Fredholm integral equation of the first kind describes the interaction of the altimeter radiation with the surface, and a constrained linear least-squares solution provides estimates of the surface's specific radar backscatter cross section {dollar}sigmasb0(phi){dollar} at incidence angles {dollar}phi{dollar} from {dollar}0spcirc{dollar} to {dollar}6spcirc{dollar}-{dollar}12spcirc{dollar} at over a million surface locations. Linear regularization minimizing the solution's second derivative ensures stability. The resolution of the observations is {dollar}0.5spcirc{dollar}-{dollar}1spcirc{dollar} in {dollar}phi{dollar} and 20-100 km in surface dimension. SAR observations at {dollar}15spcirc{dollar}-{dollar}50spcirc{dollar} yield diffuse cross section measurements. The combined scattering function estimates are organized in a spatial database at 20 km resolution.; The altimeter echo spectrum exhibits Doppler centroid shifts of up to 3 kHz from that expected from a nadir echo. These shifts cannot be attributed to navigational error, pointing error, or measurement uncertainties. Matching Doppler shifts observed by the Venera-16 spacecraft several years earlier and consistent repeat observations by Magellan of the same surface locations demonstrate a surficial origin. Specific structures consistent with the observed shifts include (i) 0.01-1 km-scale tilts of up to {dollar}1spcirc{dollar} and (ii) sudden changes in surface reflectivity {dollar}rho.{dollar} Other possible causes, such as anisotropic scattering due to 0.1-1 m-scale surface features, are postulated but not conclusively identified.; The form of the scattering function {dollar}sigmasb0(phi){dollar} varies substantially with location. Least-squares fits to three standard scattering models, Hagfors, Gaussian, and exponential, show that no single model of this type is consistent with all the observations. Overall, the exponential model fits best in plains regions and the Gaussian model fits best in complex tesserae, while the Hagfors model gives best agreement only in limited areas, typically the smoothest regions. Random surface profiles having the same height distribution and correlation functions but different slope distribution functions demonstrate that the human eye is sensitive to correlation length while the coherent radar is sensitive to slope distribution.