As said on the research list, this data is not generated on a regular basis and has a lot of fuzziness that throws off it's accuracy, but there are a couple of researchers working on the problem who may have additional ideas. I'd suggest continuing the discussion there.