Sun hours by state

A peak sun hour is one kWh of solar energy arriving on one square meter. It is a unit of energy, not of time, which is why a fourteen-hour summer day can deliver six of them and a nine-hour winter day can deliver one.

That distinction matters because the phrase sounds like daylight. Arizona and Alaska both get a long day in June; the difference is how much energy arrives during it. Tucson collects roughly twice what Anchorage collects, and the table shows that as sun hours rather than as day length.

Because it is energy per square meter, the figure converts directly into panel output. A panel rated 400 W produces 400 times the sun hours in watt hours, before system losses. That is the whole calculation, and everything else on this page is about how much the real world departs from it.

One representative city per state, plus DC. For each point, the NASA POWER climatology was queried for all-sky surface shortwave downward irradiance, which is the standard measure of what actually arrives at the ground after clouds. The annual figure is that dataset's annual value; winter is the mean of December, January and February; summer is the mean of June, July and August.

Three limits are built into that method and are worth stating plainly. The grid is half a degree, roughly 50 km across, so a value describes a region rather than a rooftop. The data is a long-term climatology, so it is not a forecast for this year. And the point chosen is one city per state, which for large states means the value may not describe where you live.

Where a state has a strong internal gradient, the chosen city is the largest population center rather than a geographic average. California shown as Los Angeles, for instance, is not a claim about San Francisco.

All 51 locations, sorted by state code. Winter and summer are three-month means, not single months, so a single cloudy season is smoothed out.

Read the winter column rather than the annual one if you are sizing. Several states sit near four sun hours on the annual figure and fall to under two in winter, which is a doubling of the array for the same winter output.

The top of the table is the desert southwest and Hawaii; the bottom is Alaska, the Pacific northwest and the upper midwest. The spread between the best and worst state is more than double, which is why a panel count without a location is meaningless.

The interesting part is not the ranking but the winter column attached to it. Phoenix at 5.8 annual still holds 3.6 in winter, while Seattle at 3.4 annual collapses to 1.3. A Seattle array sized on the annual figure is roughly three times short in December, which is the month it is needed most.

The conversion is one multiplication and one derating. Daily watt hours equal panel watts times sun hours times system efficiency. An 80 % efficiency figure covers inverter conversion, wiring loss, soiling and heat, and it is a planning value rather than a measurement of your roof.

A one kilowatt array makes the numbers easy to read: output in kWh per day is the sun hours times 0.8. The table below does that for eight states at the annual and winter figures.

Two conclusions follow. First, a small array in a good state beats a larger array in a poor one more often than people expect. Second, the winter row is the honest planning figure for anything that must work year round.

Three things move the real number far more than the state-to-state differences in this table.

Tilt and orientation. A south-facing surface at roughly latitude tilt collects close to the tabulated value. A vertical wall collects much less in summer, because the sun is high and the panel faces the horizon. A west-facing roof trades morning production for afternoon production, which sometimes suits a household better than the annual maximum.

Shade. A single shaded cell can drag a whole panel down, and a shaded panel can drag a whole string. Microinverters limit the damage to one panel, which is the main technical argument for them in small systems.

Weather in the year you are living through. The climatology is an average of decades. A cloudy winter is not a malfunction, and a bright one is not proof that the array is oversized.

For the panel-count side of this, the panel count calculator applies the same formula with your own tilt assumptions, and the appliance wattage table supplies the daily energy figure it needs.

Two or three pairs in the table repeat exactly, and that is a property of the dataset rather than an error. The climatology is a half-degree grid, so cities that fall inside the same cell receive the same value by construction. Newark and New York are one cell; Manchester and Boston are another.

The alternative would be to interpolate between cells to make the table look more varied, which would add precision without accuracy. Repeating the cell value is the honest version.

It also sets the resolution limit: differences below about 0.3 sun hours in this table are not meaningful, because they are inside the noise of a half-degree cell.

It cannot tell you what your roof collects, because that depends on pitch, azimuth, shading and the age of the glass. It cannot tell you the output of a specific panel, because panel efficiency varies by a few percent between models and more between temperatures. And it cannot tell you whether a system is worth buying, because that depends on your electricity price and your utility's export rules rather than on the sky.

Where a page on this site cites a source, that source governs; where it states an assumption, the assumption is yours to check. The method page explains how each figure is produced and how corrections are published.