We use Hapke's 1963 formulation for the reflectance function in making the synthetic models that are at the heart of converting observed earthshine intensities to terrestrial albedos. H63 has a part, at small phase angles, that correspond to the reflectances for the DS - on the BS, phase angles are large as the source is the Sun and the observer sits on Earth - but for the DS the source is the Earth. Seen from the Moon, the Earth is about 2 degrees in diameter - just what is the effective source-Moon-observer (i.e. phase) angle? For our roughly 500 observations we calculate the effective DS phase angle from geometry and a simple model of the sun-lit part of the Earth based on NCEP cloud cover, and geometry. The intensity-weighted distance between MLO and each sun-lit pixel is summed so that the effective photo-centre distance, in degrees as seen from the Moon, is arrived at. This is the plot:
Photo-centre phase angle for DS observations plotted against time of day. No correction for actual Moon-Earth distance have been performed; a standar distance of 384000 km has been used. Each linear sequence of points corresponds to an observing night, showing the evolution of phase angle as Earth rotated.
We see our observations distributed in a characteristic way: most observations early in the JD have decreasing phase angles; most observations at the end of the JD have increasing phase angles. The median phase angle is something like 1.8 degrees, with some as low as 0.5 degree and some as large as 1.6 degrees.
Since MLO is on Hawaii, opposite Greenwich in longitude, the JD change occurs for some observing nights in the middle of a set of observations - so the above plot should really be plotted from -0.5 to 0.5 instead of from 0 to 1.
We see that we are sampling H63 at reflectances near the 'opposition peak' and that the range of phase angles we have may sample enough that the errors H63 has in representing this part of the true reflectance function could be important in our work on reducing observed intensities to terrestrial albedos. We ought to have available several reflectance formalism so that we can see the importance during data-reductions. Hans has provided such additional formalism in the synthetic code, but is is complex to use (for me) so we must make an effort to get used to the new code, for the 'large paper'.