Peak Sun Hours Atlas: Every US State and European Country

A 400-watt solar panel does not make 400 watts all day. It makes the equivalent of a handful of full-power hours, and how many of those hours a site gets decides how large an off-grid array has to be. The trade calls them peak sun hours: one peak sun hour is one kilowatt-hour of sunlight landing on a square meter of ground.

This atlas gives that number for all 50 US states, the District of Columbia and 43 European countries — the annual average everyone quotes, and the December figure that quietly decides the size of the array.

What One Peak Sun Hour Is

Solar panels are rated under standard test conditions, which include 1,000 watts of light per square meter. A peak sun hour is one hour at that intensity: one kilowatt-hour of energy arriving on each square meter.

Real sunlight is weaker than that for most of the day and stronger for a short window around noon in summer, so the day’s total energy gets converted into the number of full-strength hours it would have taken to deliver the same amount.

That conversion is why the unit doubles as an energy figure. A site with 4.2 peak sun hours receives 4.2 kilowatt-hours of light energy per square meter per day.

Point a 400-watt panel at it and the panel’s theoretical harvest is 0.4 × 4.2, or 1.68 kilowatt-hours a day, before anything is lost to heat, wiring, the charge controller and the battery.

Peak sun hours are not sunshine hours. An overcast day logs no bright sunshine at a weather station and still delivers real diffuse energy to an array. This atlas counts energy, which is what a battery bank cares about.

Why December Sizes the Array

A grid-tied system runs an annual balance: summer surplus goes out to the utility, winter shortfall comes back in, and the annual average is a fair basis for sizing. An off-grid system has no such counterparty. Every kilowatt-hour used in December had to be made in December.

Batteries do not close that gap. A battery bank is sized in days of autonomy, typically two to four; a seasonal shortfall lasts eight to twelve weeks. Nothing in a normal bank stores July for January.

So the array is sized on the worst month of the year, the number designers call the design month, and the rest of the year runs in surplus by design.

In this dataset the worst month was December at every single location, from Honolulu at 21° north to Reykjavik at 64° north. That is not an assumption carried into the calculation; each site’s twelve monthly averages were compared and December came out lowest in all 94.

Tilt is the lever that partly reverses the loss. A flat panel meets December light at a punishing angle; standing the array up to roughly latitude plus 15 degrees turns it toward that low sun.

The gain is worth a median of 2.03× across all 94 locations, and it climbs with latitude: at the eighteen points above 50° north the median is 2.55×.

Here is the same 6 kWh-a-day load, sized three ways, with the 25% system losses the sizing calculator uses as its default:

Site	Sized on annual average	Sized on December, panels flat	Sized on December, panels tilted for winter
Phoenix, Arizona	1.37 kW	2.62 kW	1.39 kW
Denver, Colorado	1.66 kW	3.70 kW	1.72 kW
Madrid, Spain	1.64 kW	3.88 kW	1.79 kW
Seattle, Washington	2.32 kW	10.26 kW	4.88 kW
Berlin, Germany	2.62 kW	14.55 kW	6.25 kW

Read the last column across and the real geography of off-grid solar appears. In Phoenix, Denver and Madrid a winter-tilted array sized for December is barely larger than one sized on the annual average, because a steep tilt hands back almost everything the season took. In Seattle and Berlin it is more than twice as large.

Two things move the December number that no dataset can see. Snow on a panel produces nothing, which is a second argument for a steep tilt, since snow slides off it. And a treeline the summer sun clears easily throws shade straight across the array in December.

The Map

Both maps are colored by the same figure: December irradiation on a fixed, equator-facing array tilted to latitude plus 15 degrees, in kilowatt-hours per square meter per day. That is the design number. Use the buttons to switch continents.

United States Europe

under 0.750.75 – 1.251.25 – 2.002.00 – 3.003.00 – 4.004.00 – 5.005.00 and over

The annual averages draw a gentler map than this one. On an annual basis the continental US spans only 1.7 to 1, Arizona to Washington. In December, on a winter-tilted array, the same two states are 3.5 to 1 apart.

Europe is starker still: Cyprus and Estonia sit inside a 2-to-1 band on the annual average and nearly 10 to 1 in December.

Every US State and European Country

Rows start with the 51 US entries in alphabetical order, then the 43 European countries. Tap any column heading to re-sort, including across both continents at once. Every figure is a daily average in kilowatt-hours per square meter, which is the same thing as peak sun hours.

Annual and December, flat are horizontal-plane figures. Winter tilt is the fixed angle used: latitude plus 15 degrees, never more than 70. Tilt gain is the tilted December figure divided by the flat one. Annual ÷ December is the size of the seasonal hole.

† Alaska sits outside the coverage of the US satellite radiation product used for the other 50 rows and is computed from the reanalysis series instead.

Iceland’s December value is at the floor of what any satellite product can resolve at 64° north; both sources used here put Reykjavik below 0.05 kWh/m²/day in December, which is another way of saying that solar is not a December energy source there.

Download the full table as CSV →

The CSV adds the latitude and longitude of each reference point, the annual average on the tilted plane, and the two cross-check values described below.

How These Numbers Were Made

Each row is one point rather than a whole state or country. US rows use the largest city in the state, European rows use the capital, and that choice is named in the table and the CSV because it matters.

A large state holds more than one solar climate. Spokane, at the far side of Washington from Seattle, runs about 10% higher on the annual average and about 12% lower in December on the same winter tilt.

Monthly irradiation came from the Photovoltaic Geographical Information System run by the European Commission’s Joint Research Center. European rows use PVGIS 5.3, released in September 2024, on the CM SAF SARAH-3 satellite database covering 2005 to 2023.

US rows use the NREL National Solar Radiation Database series served by PVGIS 5.2, covering 2005 to 2015, which is the satellite product built for North America. Alaska falls outside that product’s coverage above 60° north and uses the ERA5 reanalysis series for 2005 to 2023 instead.

For every point the horizontal and tilted monthly totals were averaged across all available years, divided by the length of each month to give a daily figure, and the lowest of the twelve months taken as the worst. The tilted plane is fixed and equator-facing at latitude plus 15 degrees, rounded to the nearest degree and capped at 70, with no tracking.

PVGIS includes terrain shadowing from its elevation model; local shading from trees and buildings is beyond any dataset.

All 94 points were then pulled a second time from an independent source: the NASA POWER monthly climatology, a 20-year average covering January 2001 to December 2020, built on CERES SYN1deg satellite data by NASA Langley Research Center.

On annual horizontal irradiation the median difference between the two is 1.9%, and 93 of 94 points agree within 10%. Both values sit in the CSV, so the check is repeatable.

These figures carry no system losses. They are sunlight arriving at a surface; inverter, wiring, temperature and battery losses belong in the sizing step that follows. Nor do they model a specific roof, which has its own azimuth, pitch and horizon.

Verified September 2026. PVGIS series retrieved 2 September 2026 from PVGIS 5.3 and 5.2; NASA POWER climatology retrieved the same day from the POWER monthly climatology endpoint. Both products are updated on their own schedules; the retrieval date is the date to quote.

Common Questions

Sources: Photovoltaic Geographical Information System 5.3 (CM SAF SARAH-3, 2005–2023) and 5.2 (NREL National Solar Radiation Database, 2005–2015; ERA5, 2005–2023), Joint Research Center, European Commission. NASA POWER monthly climatology, January 2001 to December 2020 (CERES SYN1deg), NASA Langley Research Center. Standard test conditions reference irradiance of 1,000 W/m², Sandia National Laboratories PV Performance Modeling Collaborative.