Solar Irradiance Calculator (GHI / DNI / DHI → POA)
Convert GHI, DNI and DHI to plane-of-array energy. Free solar irradiance calculator using the Liu-Jordan model and NREL NSRDB state defaults.
Solar Irradiance Calculator (GHI / DNI / DHI → POA)
Site irradiance inputs
Module + economic inputs
Plane-of-array results
POA estimate uses the isotropic-sky Liu–Jordan model. It tracks NREL PVWatts v8 / PVGIS 5.2 within ±3 % for tilts ≤ 60° and azimuths within ±90° of equator-facing. Annual figures use 365 days × daily POA × PR. Specific yield = annual POA × module efficiency × PR / module area, expressed per kWp.
Show formulas and reference test
What this calculator does
Converts a site’s three irradiance components — Global Horizontal (GHI), Direct Normal (DNI) and Diffuse Horizontal (DHI), in kWh/m²/day — into Plane-of-Array (POA) irradiance for any module tilt and azimuth. POA is the single most important input in every PV energy estimate; everything downstream (annual kWh, system payback, Performance Ratio benchmarking) flows from it.
It also reports annual kWh/m², annual specific yield (kWh per kWp installed), single-module daily and annual energy, and the dollar value of one module per year at the local retail tariff. A built-in consistency check flags inputs where GHI ≠ DNI · cos(zenith) + DHI, the most common error people make when typing values out of a TMY file by hand.
How to use it
- Look up GHI, DNI and DHI for your site. The defaults match Phoenix, Arizona (NREL NSRDB 2024 typical year). For your own site, pull the values from:
- NSRDB Viewer at nsrdb.nrel.gov — pick any 4 km grid cell and download TMY3 in PSM3 format
- NREL PVWatts at pvwatts.nrel.gov — gives annual averages for your ZIP
- NASA POWER at power.larc.nasa.gov — global, free, satellite-derived
- Enter your panel tilt (0° = flat, 90° = vertical) and azimuth (180° = true south, 90° = east, 270° = west).
- Set albedo to 0.20 for typical asphalt-shingle roofs, 0.55 for fresh concrete, 0.85 for fresh snow.
- The calculator returns POA in kWh/m²/day plus annual specific yield and per-module economics.
The math, in plain English
The classic Liu–Jordan (1960) decomposition splits POA into three terms:
- Beam — what hits the panel directly from the sun:
POA_beam = DNI × cos(AOI)where AOI is the angle between the sun and the panel’s normal vector. At solar noon on the equinox, AOI ≈ |latitude − tilt| for an equator-facing array. - Sky diffuse — scattered light from the sky dome:
POA_diffuse = DHI × (1 + cos β) / 2. A flat panel (β = 0) sees the full sky dome; a vertical panel (β = 90°) sees only half. - Ground-reflected — light that bounces off the ground:
POA_ground = GHI × ρ × (1 − cos β) / 2. Higher tilt and brighter ground (snow) increase this term.
The total POA = beam + diffuse + ground is then multiplied by 365 to get annual kWh/m², and by module efficiency × PR × area for per-module energy.
Per-state irradiance, NREL NSRDB 2024
NREL’s National Solar Radiation Database is the authoritative US reference. Annual daily-average GHI varies by a factor of 1.7× across the lower 48 — Seattle gets 3.4 kWh/m²/day, Yuma AZ gets 6.5. DNI varies even more (2× range) because high-DNI desert sites lose less to clouds.
| Region | Typical city | GHI (kWh/m²/day) | DNI (kWh/m²/day) | DHI (kWh/m²/day) |
|---|---|---|---|---|
| Pacific NW | Seattle WA | 3.40 | 2.95 | 1.55 |
| Northeast | Boston MA | 4.05 | 4.10 | 1.70 |
| Midwest | Chicago IL | 4.10 | 4.05 | 1.75 |
| Southeast | Atlanta GA | 4.85 | 4.95 | 1.80 |
| South | Austin TX | 5.10 | 5.65 | 1.75 |
| Mountain West | Denver CO | 5.25 | 6.30 | 1.65 |
| Desert SW | Phoenix AZ | 5.79 | 7.29 | 1.71 |
| Hawaii | Honolulu HI | 5.95 | 5.80 | 2.05 |
Source: NREL PSM3 TMY3, 1998–2022 baseline, accessed 2024 Q4.
What POA tells you about system sizing
Once you know annual POA in kWh/m²/year, the design chain is straightforward:
- Annual specific yield (kWh per kWp DC) = annual POA × PR. A south-facing 30° array in Phoenix with PR 0.78 yields ≈ 6.91 × 365 × 0.78 ≈ 1968 kWh/kWp, which matches NREL PVWatts v8 within 2 %.
- System size for a given annual kWh demand:
kWp = annual_kWh / specific_yield. A 12,000 kWh Phoenix household needs ≈ 6.1 kWp. - Number of panels =
kWp / panel_kWp. At 400 W panels, that is 16 modules; at 440 W, 14 modules. Cross-check with the solar panel count calculator which folds in roof-area constraints.
Practical tips for accuracy
- Always use TMY data, not a single year. A single low-irradiance year (volcanic dust, El Niño) can underpredict 25-year fleet output by 10 %. NSRDB TMY3 averages the most representative month from each calendar month across a 20+ year baseline.
- Update albedo seasonally if you are above 40°N. Summer asphalt is 0.18; January snow cover takes the local albedo to 0.55–0.85, lifting POA on a 60°-tilt panel by 8–15 % during the heating-load months. The solar snow loss calculator handles the loss side; this calculator handles the reflectance gain.
- For bifacial modules, double-count the rear face. A typical bifacial gain at 0.20 albedo and 1 m clearance is 5–8 % over an equivalent monofacial array. Multiply the ground-reflected term by 1.15–1.20 as a first approximation.
- Cross-check against the NREL PVWatts Calculator at pvwatts.nrel.gov before finalising a design. PVWatts uses Perez transposition and a richer thermal model; expect agreement within ±3 % with this calculator at residential tilts.
How POA feeds the rest of your design
POA is the upstream variable for almost every other calculator on this site:
- The solar panel output calculator takes POA × PR as its core energy estimate.
- The solar system efficiency calculator inverts the relationship — given measured AC output and POA, it returns a real-world PR you can benchmark against IEC 61724-1 typical ranges.
- The solar panel tilt calculator and solar panel azimuth calculator feed directly into the AOI term of the beam component.
- The cost of solar panels calculator divides total system cost by lifetime energy (annual POA × 25 × system kWp × PR) to express LCOE in $/kWh.
When to use a more sophisticated model
The isotropic Liu–Jordan transposition used here is a deliberate simplification. NREL’s full PVWatts engine, SAM (System Advisor Model), PVsyst and Helioscope all use the Perez (1990) transposition, which adds a circumsolar and horizon-brightening correction. The Perez model is 1–3 percentage points better at predicting POA on south-facing arrays at moderate tilts and 3–5 points better for east/west arrays at steep tilts.
For preliminary sizing, ROI estimates and tilt/azimuth sensitivity studies, the isotropic model is the right tool — it is fast, transparent, and accurate to within the noise of the underlying TMY data. For final commissioning paperwork, contractual yield guarantees, or any utility-scale project above 1 MWp, run the numbers through PVsyst or SAM with hourly Perez transposition and a site-specific TMY.
The NREL Solar Resource Best Practices Handbook (2024 edition) is the definitive open-access reference for understanding when each transposition model is appropriate and how much accuracy you give up by simplifying.
Frequently asked questions
What is the difference between GHI, DNI and DHI?
What is POA irradiance and why does it matter?
Where can I get GHI, DNI and DHI data for my site?
What is a typical Performance Ratio (PR) for a residential PV system?
Why does the calculator use an isotropic sky model?
Related calculators
📋 Embed this calculator on your site (free, attribution required)
Free to embed on any non-commercial or commercial site, provided the attribution link remains visible. No tracking, no email capture, just the calculator.