So here is a commutative plot of all the data logged in Operation Lux so far, some 21 days worth. The dots are individual data points and they are colored by day. The solid black line is an average plot of all the data from all the days, where the red plot is an average of only the top quarter of the data. This shows the general shape of the light curve throughout the day. There are a couple of artifacts specific to the location on the roof the experiment sits. Notice how around 9:00 am there is a sudden spike in light. This is the time when the sun rises over the tree line. It sets behind the trees again around 5:30 pm. With all the data collected so far we can say that during the summer months this portion of the roof can expect an average of around 3,100 watt hours per meter square of sunlight in a 24 hour period. To run the Raspberry Pi off sunlight alone on an average day would require a 30 watt solar panel, and an 8 amp hour battery. Cloudy days will require an external power source unless a larger battery and/or solar panel is used.
The days are getting shorter as summer draws on, but the fall equinox is not until late September. It will be interesting to see how the light is effected as the sun gets lower in the sky, especially with our fairly treeline. We could try some theoretical approaches to making this estimate. For example, the sunrise for today in Middleton, WI happened at 5:56 am and the sun set at 8:19 pm for a total of 14 hours, 14 minutes of daylight. The winter solstice will produce 8 hours, 59 minutes of light—37% less light time than today. By scaling the light curve to force it into this shorter amount of time will produce a number that would reflect the energy available from the shorter day. However, it will not take into account the loss of light from the sun's position in the sky (it will never get as high). So having the Operation Lux take the measurements is still the best way to get this data.
