Accurate modeling of acoustic propagation in the ocean waveguide is important for SONAR performance prediction, and requires, among other things, characterizing the reflection properties of the bottom. Recent advances in the technology of autonomous underwater vehicles (AUV) make it possible to envision a survey tool for seabed characterization composed of a short array mounted on an AUV. The bottom power reflection coefficient (and the related reflection loss) can be estimated passively by beamforming the naturally occurring marine ambient-noise acoustic field recorded by a vertical line array of hydrophones. However, the reduced array lengths required by AUV deployment can hinder the process, due to the inherently poor angular resolution. In this paper, data from higher frequencies are used to estimate the noise spatial coherence function at a lower frequency for sensor spacing beyond the physical length of the array. This results in higher angular resolution of the bottom loss estimate, while exploiting the large bandwidth available to current acquisition systems more efficiently than beamforming does. The technique, rigorously justified for a halfspace bottom, proves to be effective also on more complex bottom types, both in simulation and on experimental data.