Solar Insolation: GHI, DNI & Peak Sun Hours

Insolation drives the energy estimate behind every PV design, proposal and financial model. Module efficiency, cabling and soiling all matter, but none of them rescues a wrong insolation assumption. As a first-order rule, a 10% error in the irradiation input is about a 10% error in annual yield. On a 500 kWp roof with an assumed 1,400 kWh/kWp specific yield, that is 70,000 kWh a year of misestimated production.

Last updated: November 2026

SurgePV publishes this post and sells solar design software. The data tables below were calculated with the free PVGIS tool, so you can reproduce them.

This article is a designer’s reference. Each section ties insolation data to a decision in design, sizing or proposal writing: the metrics, the arithmetic, the data sources and the mistakes that make estimates miss.

In this guide:

• What solar insolation means in designer terms
• Irradiance versus insolation, and why STC uses 1,000 W/m²
• GHI, DNI and DHI, and which one each technology uses
• Peak sun hours and the step-by-step sizing formula
• Tilt angle optimization with PVGIS-calculated city data
• Solar constant, air mass and why a module rarely makes its nameplate power
• Data sources designers use
• How shading and soiling reduce insolation before it reaches the cell
• Common mistakes in insolation assumptions
• Quick-reference tables for US and international cities

What Is Solar Insolation? (The Designer’s Definition)

Solar insolation is the solar energy received on a surface over a defined period, measured in kilowatt-hours per square metre (kWh/m²). For designers the practical unit is kWh/m² per day, because multiplying it by system size gives a first estimate of daily energy. Annual values in kWh/m² per year suit feasibility and finance work.

Sandia’s PV Performance Modeling Collaborative states the relationship simply: irradiance is to power as insolation is to energy. Its units are watts per square metre for irradiance and kilowatt-hours per square metre for insolation, as in the PVPMC irradiance and insolation reference. The site also has glossary entries for insolation and solar irradiance.

The solar constant is the irradiance at the top of Earth’s atmosphere on a surface facing the sun. NASA’s Goddard solar irradiance page says SORCE established a value of 1361 W/m², and that TSIS-1 gives 1361.6 ± 0.3 W/m² for the 2019 solar minimum. Atmosphere, cloud and sun angle reduce that to much lower values at the ground. See NASA’s solar irradiance science page.

Daylight duration alone tells you little. Cairo in December has roughly ten hours of daylight and London in December has under eight, yet PVGIS gives 3.43 kWh/m² a day of horizontal irradiation for Cairo against 0.67 for London. Duration is one input among sun height, atmosphere and cloud. For comparison, London in July is 5.14 kWh/m² a day in the same tool.

This is why designers size with peak sun hours (PSH) instead of daylight hours. One peak sun hour equals 1 kWh/m². A location with 5.2 peak sun hours a day receives 5.2 kWh/m² a day, and multiplying that by system size and a derate factor gives a first production estimate. The glossary has a page on peak sun hours.

The atmospheric path length, expressed as air mass (AM), explains why ground-level irradiance is below the solar constant. A simple approximation is air mass equal to one divided by the cosine of the solar zenith angle. At the equator at solar noon on an equinox the sun is near the zenith and air mass is close to 1. At 40° latitude at winter solstice noon the zenith angle is about 63°, which gives an air mass above 2. More atmosphere means more scattering and absorption, which is one reason winter insolation falls faster than day length alone predicts.

Always use location-specific data. Do not extrapolate from a nearby city without checking latitude, elevation and cloud patterns. A coastal site with a morning marine layer can differ from an inland site a short drive away, and the dataset’s resolution decides how much of that it can resolve.

Irradiance vs. Insolation, and Why STC Uses 1,000 W/m²

Irradiance is instantaneous power density in W/m². Insolation is cumulative energy in kWh/m². A car at 80 km/h for 3 hours covers 240 km. A roof receiving 800 W/m² for 5 hours accumulates 4.0 kWh/m².

The 1,000 W/m² on a module datasheet is a reference, not an average. PV Education states that testers reproduce the standard test conditions (STC) of 1000 W/m² of sunlight, 25 °C and a spectrum of air mass 1.5, per its module measurement page. The reference spectrum at air mass 1.5 corresponds to a solar zenith angle of 48.19°, according to the National Laboratory of the Rockies AM1.5 reference spectra page. The exact STC definitions sit in the IEC 60904 series, so check the current standard if a contract cites it. See also the glossary entry on standard test conditions.

STC is a laboratory benchmark. It lets you compare modules on one basis, but it is not a promise of field output. Cell temperature, irradiance level and spectrum all differ outdoors, and the module’s temperature coefficient sets how much power falls when cells run hot.

As an illustration with assumed numbers, take a 400 W module with a datasheet coefficient of -0.35% per °C. At a 55 °C cell temperature, 30 °C above 25 °C, power falls by about 10.5% (30 x 0.35%). At 900 W/m² instead of 1,000, output scales roughly in proportion. That gives about 400 x 0.895 x 0.9, or 322 W, before wiring, mismatch and inverter losses. The point is the method, so use your module’s own datasheet.

GHI, DNI, and DHI: The Three Components

Solar radiation at the ground splits into measurable components. Knowing which one matters for your technology avoids mismatches.

Glossary pages cover GHI, DNI and DHI. Sandia’s PVPMC page notes that direct irradiance is measured with a 5° field of view, while a 180° field of view is sometimes called global.

On a horizontal surface the relationship is:

GHI = DHI + (DNI x cos(θz))

Here θz is the solar zenith angle. As an example with assumed values, at an equinox noon at 40° latitude the zenith angle equals the latitude, so cos(40°) is 0.766. If DNI is 850 W/m² and DHI is 150 W/m², GHI is 150 + 850 x 0.766, or 801 W/m².

Two sites can have very different annual GHI. In a PVGIS run for this guide, Phoenix gave 2,145 kWh/m² a year of horizontal irradiation and Berlin gave 1,115. A fixed-tilt flat-plate PV system works at both. A concentrating solar power (CSP) plant needs direct beam radiation, so it is only worth evaluating where DNI is high, and the developer’s own threshold governs that decision.

For flat-plate PV, GHI is the starting point. Design software splits GHI into direct and diffuse parts, transposes each onto the tilted array plane and sums them into plane-of-array (POA) irradiance, a term covered in the glossary under plane-of-array irradiance. The choice of transposition model affects results, especially in cloudy climates where diffuse light dominates, so check which model your tool uses.

Peak Sun Hours and the Sizing Formula

Peak sun hours turn insolation into module counts. A site with 5.2 PSH a day receives the same energy as 5.2 hours at 1,000 W/m², and that linearity is what makes the formula work.

Here is the walkthrough from utility bill to module count, using assumed inputs.

Step 1: Find daily energy use. A home uses 500 kWh a month. Divided by 30 days, that is 16.7 kWh a day.

Step 2: Look up peak sun hours. Assume the site has 5.2 PSH a day. Use a documented source such as PVGIS or NSRDB, and record the dataset and year range.

Step 3: Apply a derate factor. Real systems lose energy to temperature, soiling, shading, mismatch, wiring and conversion. This example assumes 0.84, which is a planning assumption and not a tool default. Replace it with the loss model in your simulation tool, and document the reason for any change.

Step 4: Calculate DC size.

PV system size (kW) = daily kWh ÷ peak sun hours ÷ derate factor

16.7 ÷ 5.2 ÷ 0.84 = 3.82 kW DC.

Step 5: Convert to modules. With 440 W modules, 3,820 W ÷ 440 W is 8.7, rounded up to 9 modules, or 3.96 kW DC. In practice you also round to fit roof geometry, inverter voltage windows and fire setbacks.

This is the backbone of a preliminary proposal. It does not replace hourly simulation. The derate factor is where hand calculations drift. Using 0.90 instead of 0.84 undersizes the array by about 7%, and using 0.75 instead of 0.84 oversizes it by about 12%, because the system size scales with one over the derate factor. Document the assumption in every proposal, for example “0.84 planning assumption, adjusted to 0.80 for documented tree shading on the southwest roof face.”

Peak sun hours vary by season, and that matters for production estimates. In a PVGIS run for Denver, horizontal irradiation was 7.26 kWh/m² a day in July and 2.47 in December. Annual averages suit payback screening, but a client who cares about winter self-consumption or export limits needs the monthly view. PVGIS and the NLR PVWatts Calculator both give monthly output.

Tilt Angle Optimization

Tilt is a design parameter installers control, yet many flat commercial roofs ignore its energy cost. The table below shows PVGIS results for fixed-tilt arrays: the tilt that maximizes annual energy and how much a flat array loses against it.

Source: PVGIS 5.3 PVcalc, 1 kWp, 14% system losses, terrain horizon off, queried November 2026. Penalty is the annual energy of a 0° tilt array compared with the optimal-tilt array at the same site.

The rule of thumb “tilt equals latitude” works near the tropics but overshoots at high latitudes in this data. At 47.6°N, Seattle’s best fixed tilt was 40°. At 52.5°N, Berlin’s was also 40°. Treat latitude as a starting value and let the simulation choose.

How sensitive is yield to tilt? For Denver in the same tool, an array at 30° or 50° produced about 1.2% to 1.3% less annual energy than 40°, and a 20° tilt produced about 4.9% less. Flat mounting produced about 19.6% less. Azimuth matters as well, so run the actual orientation of each roof plane in your simulation rather than assuming a rule.

The commercial-roof lesson is practical. Flat mounting is often driven by structural loading, wind uplift or ballast limits, not energy. When you propose a flat array, quantify the production loss and show the tilted alternative with its structural trade-off. Let the client choose with data. For a residential walkthrough, see how to design a residential solar system, and for the glossary view see solar panel tilt angle.

Solar Constant, Air Mass, and Real-World Output

The solar constant is about 1361 W/m² outside the atmosphere, per NASA. That is the theoretical ceiling. Ground-level insolation is what remains after atmospheric absorption, scattering and cloud.

AM0 refers to the spectrum outside the atmosphere, used for space PV. It matters to rooftop designers only as a reference. AM1.5 is the standard ground-level reference spectrum for module testing. Its zenith angle is 48.19° according to the NLR reference spectra page.

Air mass depends on sun height. Close to the equator at noon the air mass is near 1. In winter at high latitudes the path through the atmosphere is several times longer, which lowers irradiance sharply. That is part of why insolation drops so much in winter at high latitudes: the days are shorter and the path is longer.

Soiling adds another real-world loss. Dust, pollen, bird droppings and particulates reduce the light reaching cells, and satellite-derived insolation databases assume a clean surface. Soiling is site-specific, so measure it. The glossary covers soiling loss, and your production data shows whether a cleaning schedule pays.

Data Sources Designers Use

Not every solar resource source suits every project. Match the dataset to geography, scale and financing needs, and record the dataset name, version and period in every yield report.

This table makes no accuracy claims, because the pages reviewed for this guide did not publish comparable figures. Check each tool’s documentation for resolution and validation before relying on it.

The JRC PVGIS page describes the tool as providing solar radiation and PV performance information for any location worldwide, with maps, typical meteorological year data and APIs. For U.S. designs, the NSRDB site is the usual entry point.

For bankable reports, lenders and independent engineers often specify the dataset they accept, and commercial datasets are common. Ask them which one they require before you commission modeling, and cross-check at least two sources on large projects. If they disagree materially, find out why before you present numbers to a client or lender.

How Shading and Soiling Reduce Insolation

Shading and soiling are the two largest site-specific reductions that regional datasets cannot see. A tree that blocks the morning sun, a neighboring building casting an afternoon shadow or weeks of dust each reduce the energy reaching the module.

Production loss from shading can exceed the shaded fraction of insolation. Shaded cells in a string limit current for the whole string, and bypass diodes limit the damage without removing it. The size of the effect depends on string layout, module-level electronics and the shading pattern, so model it per site. Glossary pages cover shading analysis and the wider workflow.

Total solar resource fraction (TSRF) is a common way to express the combined effect of shading and orientation. It is the product of solar access and the tilt and orientation factor (TOF):

TSRF = solar access x TOF

Solar access is the ratio of insolation reaching the array to what an unshaded array at the same tilt and azimuth would receive. TOF compares insolation at the actual tilt and azimuth with the optimum. With assumed values of 92% solar access and 98% TOF, TSRF is 0.92 x 0.98, or 0.9016, which is about 90.2%.

Soiling is invisible in design software unless you apply a derate. Climate, rain pattern, tilt and nearby sources decide the loss, so use measured data from comparable sites or from your own monitoring where you can.

Three mistakes recur: using default soiling where the climate is dusty, using unshaded GHI for a roof surrounded by mature trees, and applying a single TSRF to the whole array when shading affects one string. The fix is site-specific shading analysis with hourly obstruction geometry over the full year.

Common Mistakes That Blow Up Proposals

Insolation errors follow a project into commissioning and operations. These six are the ones to check first.

1. Using annual average insolation for a system sized against worst-month demand. An annual average can look healthy while the winter month is far lower. For Denver in PVGIS, December horizontal irradiation was about one third of July’s. Size for the constraint the customer cares about, then test whether the array still makes financial sense.

2. Applying horizontal GHI to a tilted array without transposition. Databases report GHI, while the array sees POA irradiance. For Denver in PVGIS, annual in-plane irradiation at a 40° tilt was 2,311 kWh/m² against 1,886 kWh/m² horizontal, about 22% higher. Using raw GHI as the array input would understate yield. Use software that transposes automatically, and check the transposition model.

3. Ignoring soiling derates in dusty regions. A default allowance suited to rainy zones can be too low in arid or industrial areas. If a performance guarantee applies, a low soiling assumption can trigger a shortfall.

4. Using city-average data for microclimate sites. Fog belts, coastal influence and urban heat can make a site differ from the grid cell around it. If the site is far from the data point or sits in a known microclimate, widen your uncertainty band and verify against local measurements.

5. Confusing DNI with GHI for flat-plate module counts. A high DNI figure from a CSP dataset does not size a flat-plate array. Diffuse light counts for flat-plate PV, so use GHI or POA irradiance.

6. Promising production on an open horizon without shading analysis. Unshaded GHI assumes no obstructions. A tall building to the west or a stand of trees will take insolation, and the year-one shortfall will show in the customer’s first bills.

Quick-Reference Tables for Designers

These values were calculated for this guide with the JRC’s PVGIS 5.3 PVcalc tool in November 2026: 1 kWp, 14% system losses, terrain horizon off, fixed tilt at the PVGIS-optimized angle (north-facing in the southern hemisphere). Radiation data come from PVGIS-ERA5 for most locations and PVGIS-SARAH3 for London, Berlin and Dubai. Peak sun hours here means annual horizontal irradiation divided by 365. Use them for orientation, not for a final design.

US Cities

International Cities

The spread is wide. London averages 2.93 peak sun hours on this basis and Dubai 5.95, about twice as much. Identical system sizes in those two places would produce very different energy, so size from location data every time.

For final design, pull site-specific data from a documented source and have the same inputs reproducible by a reviewer. The tables above are starting points.

How SurgePV Fits an Insolation Workflow

SurgePV is solar design and proposal software. Its published scope covers 3D roof modeling, array layout, shading analysis, energy-yield modeling, financial modeling and proposal generation, so an insolation assumption can be carried from roof model to customer document in one workflow. The platform does not publish which solar resource dataset it uses, so ask for that in a demo and record it with every yield report.

1. Roof model. Build the roof planes, pitch, azimuth and obstructions in solar design software.
2. Shading. Run shading analysis on the 3D model to see obstruction losses by hour and season.
3. Energy and financials. Use the generation and financial tool to apply losses and see production and returns.
4. Proposal. Turn the results into a customer document with solar proposal software.

Results depend on source data, assumptions, equipment models and review, and outputs do not replace approval by the responsible engineer, authority, lender, insurer or utility. Time a real project against your current process before you decide, and see the pricing page for plans.

Conclusion

Insolation is the foundation of PV design. Module counts, string sizing, energy yield, payback and PPA pricing all start from it. A designer who handles GHI, DNI, DHI, peak sun hours, tilt and shading corrections carefully produces estimates that survive commissioning.

Three actions to take from this reference:

1. Audit your insolation source. Record the dataset, version and period in every yield report, and cross-check a second source on commercial projects.
2. Add shading and soiling to the standard derate checklist. Regional data is a starting point. Walk the site, model obstructions in 3D and apply a soiling factor that suits the climate.
3. Document every assumption in the proposal. Data source, tilt, azimuth, derate factor and shading loss. A proposal with documented assumptions is defensible, and a single yield number is not.

To see how insolation inputs flow from roof model to proposal, book a demo.

Frequently Asked Questions

What is the difference between solar irradiance and solar insolation?

Solar irradiance is instantaneous power, measured in watts per square metre (W/m²). Solar insolation is energy accumulated over time, usually in kilowatt-hours per square metre (kWh/m²). Sandia’s PV Performance Modeling Collaborative puts it this way: irradiance is to power as insolation is to energy.

How do you calculate solar insolation for a PV system?

Look up daily peak sun hours for the site from a solar resource tool such as PVGIS or NSRDB. Divide daily energy use by peak sun hours and by a derate factor for system losses to get DC size. For example, 16.7 kWh a day, 5.2 peak sun hours and an assumed 0.84 derate gives about 3.8 kW.

What is a peak sun hour?

One peak sun hour is 1 kWh/m² of solar energy, the same as one hour at 1,000 W/m². A site with 5.2 peak sun hours a day receives 5.2 kWh/m² a day. Designers use it to turn insolation data into a first estimate of energy yield.

What is the best tilt angle for solar panels?

It depends on latitude and climate. In a PVGIS run for 17 cities, the best fixed tilt for annual energy was near latitude in sunny low-latitude cities such as Miami and Honolulu and well below it at high latitudes, for example 40 degrees at 52.5 degrees north in Berlin. Check your own site and compare seasonal needs.

Which solar insolation database is most accurate?

No source is best everywhere, and none replaces site checks. Use a tool that covers your region, such as PVGIS for much of the world or NSRDB for U.S. designs, and compare at least two sources on large projects. Ask a lender or independent engineer which dataset they require.

How does shading affect solar insolation?

Shading cuts the share of available insolation that reaches the modules, and the production loss can exceed the shaded fraction because shaded cells limit the current of the string. Model obstructions on a 3D site model, then check the result with a site visit.

What are standard test conditions for solar panels?

Standard test conditions (STC) are 1,000 W/m² of irradiance, a 25 °C cell temperature and an air mass 1.5 spectrum, per PV Education. Module power ratings are given at STC, so real output differs with temperature, irradiance and spectrum.