Peak sun hours are almost always quoted as an annual average, and the average is the right input for annual sizing. It also conceals a swing of roughly three to one between midwinter and midsummer, and nearly every disagreement about system size is really a disagreement about which season you are sizing for.
The sizing formula is in sizing a home solar array.
Daily output per kW installed
daily kWh = system kW × peak sun hours × derate
At a derate of 0.80:
| Region | Sun hours | kWh/day per kW | kWh/year per kW |
|---|---|---|---|
| Pacific Northwest, interior New England | 3.5 | 2.8 | 1,022 |
| Great Lakes, Upper Midwest | 4.0 | 3.2 | 1,168 |
| Mid-Atlantic, Midwest, New England coast | 4.5 | 3.6 | 1,314 |
| Mid-South, Central Plains | 5.0 | 4.0 | 1,460 |
| Southwest, Southern California | 5.5 | 4.4 | 1,606 |
A 10 kW system therefore produces roughly 10,200 kWh a year in the Pacific Northwest and 16,100 in the Southwest. Same equipment, same cost, 57 % more energy.
These are regional bands, not state figures. Several large states span two or more bands — elevation, coastal fog and mountain shadow all matter more than the state line. For a specific address, model it rather than looking up a state.
The seasonal swing
The annual average hides the shape. Roughly, for a fixed south-facing array in the middle latitudes of the US:
| Season | Relative to annual average |
|---|---|
| Midsummer | ~1.4× |
| Spring and autumn | ~1.0× |
| Midwinter | ~0.5× |
Midsummer produces close to three times midwinter. Two effects compound to produce that:
The sun is lower in winter, so the same energy is spread across more atmosphere and arrives at a shallower angle to a fixed panel. A panel tilted for the annual average is badly angled for December.
The days are shorter. Fewer hours of any sunlight at all.
There is a partial offset — panels are more efficient when cold, so a clear January day produces better per hour of sun than a hot July one. It does not come close to closing the gap.
Why this matters for sizing
If your utility offers annual net metering, the swing largely does not matter: summer surplus banks against winter deficit and the annual total is what counts. Size to the annual figure and stop thinking about it.
If your utility settles monthly, or credits exports below retail, the swing matters a great deal. Summer surplus is worth less than winter shortfall costs, so the economically optimal array is smaller than the annual-total calculation gives, and shaped around your consumption pattern rather than your total.
A house with electric heat has the worst possible alignment: peak consumption in January, minimum production in January. A house whose load peaks with summer air conditioning has the best.
Tilt and orientation
Two adjustments that move the number:
Tilt. A tilt roughly equal to your latitude is a reasonable annual optimum. Steeper favours winter, shallower favours summer. In practice most residential arrays sit at whatever the roof pitch is, and the difference between roof pitch and optimum is usually a few percent — not worth mounting hardware to correct.
Azimuth. True south is the reference in the northern hemisphere. East or west facing loses roughly 10–20 % annually depending on latitude, and shifts the production curve into the morning or afternoon. On a time-of-use tariff a west-facing array can be worth more than a south-facing one despite producing less, because it produces into the expensive late-afternoon window.
Where the derate goes
Every figure above uses 0.80, which is a reasonable default and not a law. It is the product of inverter efficiency, wiring losses, soiling, temperature, orientation and panel tolerance — broken down on the main sizing page.
Adjust it. A clean, well-ventilated, well-oriented array can justify 0.84. A flat-mounted array on a hot dark roof with afternoon shade cannot.
What this page does not cover
- Shading, which is site-specific and not a percentage.
- Snow cover, which in northern climates can take out days of winter production entirely and is not in any sun-hour figure.
- Tracking mounts, which follow the sun and produce more but are rare on residential roofs.
- Hourly modelling. Tools that model real weather for your coordinates will beat any regional band.
Annual averages for sizing, seasonal shape for arguing about it.