The Full Forward Method for determining not only earthshine intensity but also terrestrial albedo has been tested on, so far, a singe synthetic lunar image corresponding to a few days after New Moon.
By minimizing on the difference between the ‘observed image’ and trial models from Hans’ synthetic image generator over the whole image frame, for three values of PSF width, we can retrieve the terrestrial albedo to within 0.07% for alfa=1.6, 1.7 and 1.8.
This is below our science requirement, and gives confidence!
The model must be tested for images corresponding to a whole month now. The calculation is time-consuming since it is not programmed for the DMI Cray yet, so meanwhile the method runs in a mixture of Fortran and IDL on a workstation.
The fitting method can handle offsets in the image so to some extent an incorrect bias subtraction should be subsumed by this analysis method.
The method relies on Chris’ insight that a library of lunar images for a given phase but different terrestrial albedos can be generated from the linear sum of an ‘albedo is 0’ image and an ‘albedo is 1’ image. The weighting function between these extremes is simply the albedo and is one of the fitting parameters.
Testing remains also for how dependent we are on knowing the PSF – i.e. not just the exponent but also things like non-power-law shape.
This result brings to an end many weeks of worries I had that we would not in time be able to show that we have a method that is good, and now – after some more calculations overnight – we should be on line to write the results section of our paper!