I'm returning to a project that I started working on about 2 years ago, and see that a question was asked about cloud cover in the 3-hourly 5-day Site Specific Forecast. That would probably help me too, but what I am actually trying to do is derive a prediction around midnight of the likely total insolation for the next day. I've tried a number of combinations of UV, weather type, and visibility, but can't help thinking that all I'm doing is reverse engineering what you did to generate those forecast elements in the first place. Are there any plans to include either % cloud cover or better still, hours and strength of sunshine into the 3hourly?