How reliable is the weather in a location? This has to be one of the primary considerations when setting up a solar venture. The length of the daylight period is known and predictable from year to year. What is less predictable is the amount of cloud and other atmospheric content which can attenuate the amount of solar radiation.
According to NASA(1) the sun is very constant in its light output. However the light has to penetrate layers of atmosphere which can vary enormously in content that can block this light. Moisture, clouds, dust, particulates all have a role in either reflecting light back out into space or absorbing the light energy so that it never has the chance to excite the electrons in the solar array and put energy into the grid. Temperature also has a role in that the hotter the panels the less efficient they are in generating power from any given level of light input.
Clearly it matters where the solar array is located on the earth's surface. In Australia and Arizona the amount of sunlight reaching the earth in those dry climates is a lot higher in an average daylight period than in the maritime and frequently cloudy skies of Newfoundland.
My array in South Eastern Ontario has been generating for five full years now. The monthly numbers tell as story, and a fairly constant picture it is. Here are some numbers for each month of the five year period. For each month I have calculated the range (the largest monthly generation less the smallest) and then standardized this by dividing by the mean for that month over the five years. Standardization is helpful since the generation numbers for mid-summer are about double those for mid-winter.
Month Range/Mean January 0.42 February 0.27 March 0.17 April 0.25 May 0.25 June 0.20 July 0.19 August 0.14 September 0.10 October 0.35 November 0.37 December 0.16 Year 0.08
The largest variation is in January, when the range was quite large compared to the mean, and the smallest is in September when the monthly numbers were fairly consistent. In fact we see that in general the summer numbers are more steady than those for the winter, despite the fact that cooler temperatures will make the generation more efficient. As the number for the year shows, the year to year production levels are remarkably similar, which is probably a result of the long-day generation in summer being so steady.
One conclusion is that provided the weather patterns remain normal for the region, the yearly production will be reliable; production lost in one month will likely be restored in another. Therefore the uncertainty lies in the probability of things not being normal. What are the chances that a major volcanic eruption will fill the air with dust particles, or that an El Nino event will cast more moisture into the atmosphere? And what are the chances that the variability of generation in summer increases?