Every solar yield estimate begins with a resource number: how much sunlight reaches the site, in what form, and from which direction. Three irradiance components — Global Horizontal Irradiance (GHI), Direct Normal Irradiance (DNI), and Diffuse Horizontal Irradiance (DHI) — describe different aspects of that resource. Choosing the right one for your system type is not optional: using DNI where GHI is appropriate, or vice versa, can produce yield errors of 10–30% before a single module is installed.
The three components of solar irradiance
Global Horizontal Irradiance (GHI) is the total solar radiation received on a horizontal surface. It is the sum of direct beam radiation projected onto a horizontal plane plus all diffuse sky radiation. GHI is measured by a pyranometer mounted flat. It is the most widely available irradiance dataset and the standard starting point for fixed-tilt PV yield modelling.
Direct Normal Irradiance (DNI) is the solar radiation received on a surface that is always perpendicular to the sun — i.e., it tracks the sun through the day. DNI measures only the direct beam component, with no diffuse contribution. It is measured by a pyrheliometer mounted on a sun tracker. DNI is the relevant resource metric for concentrating systems (CSP, CPV) and is the component maximised by single-axis and dual-axis solar trackers.
Diffuse Horizontal Irradiance (DHI) is the solar radiation received from the sky dome excluding the direct beam — scattered by clouds, aerosols, and air molecules. It is measured by a pyranometer with a shading disc blocking the direct beam. In cloudy climates, DHI can represent 40–60% of GHI. In clear desert conditions, it may drop to 10–15%.
The relationship between the three is: GHI = DNI × cos(θ) + DHI, where θ is the solar zenith angle (angle between the sun and vertical). This decomposition identity is fundamental to all transposition calculations.
Which component to use for different system types
| System Type | Primary Resource Metric | Why |
|---|---|---|
| Fixed-tilt PV (rooftop or ground mount) | GHI | Panels receive both beam and diffuse radiation; GHI is the starting point for transposition to tilted surface (POA) |
| Single-axis tracker (SAT) | GHI (or GTI modelled) | Yield tools transpose GHI to rotating-plane irradiance; high DNI locations benefit most from tracking |
| Dual-axis tracker | DNI + DHI | Dual-axis follows the sun perfectly; yield = DNI + DHI component on normal surface |
| Concentrating Solar Power (CSP) | DNI only | Mirrors concentrate direct beam only; diffuse radiation cannot be focused |
| Bifacial fixed-tilt | GHI + albedo | Front uses transposed GHI; rear captures diffuse + ground-reflected radiation |
| Building-integrated PV (BIPV) | GTI (plane-of-array) | Vertical or angled surfaces require full transposition including circumsolar diffuse |
Irradiance data sources: what to use for project development
Several authoritative databases provide long-term monthly and hourly irradiance data:
- NASA POWER: free, global coverage, hourly data back to 1984. Resolution ~50 km. Best for early-stage screening.
- PVGIS (EU Joint Research Centre): free, excellent coverage of Europe, Africa, and Asia. Resolution ~5 km. Standard for European projects.
- SolarAnywhere / Solargis: commercial products with resolution down to 250 m, P50/P90 uncertainty analysis, and long-term trend data. Required for bankable yield reports on projects >500 kWp.
- Meteonorm: industry-standard TMY (Typical Meteorological Year) generator used directly in PVsyst. Combines satellite and ground-station data.
- Ground station pyranometers: most accurate but require 1–3 years of on-site measurement. Mandated by some lenders for projects >10 MWp or in data-sparse regions (parts of Central Asia, Sub-Saharan Africa).
GHI maps and regional benchmarks
| Region | Annual GHI (kWh/m²/yr) | DNI (kWh/m²/yr) | Tracker Benefit |
|---|---|---|---|
| Northern Europe (Germany, Poland) | 950–1,200 | 700–950 | Low (10–15%) |
| Southern Europe (Spain, Italy) | 1,500–1,900 | 1,400–1,800 | Moderate (18–22%) |
| Middle East (UAE, Saudi Arabia) | 1,900–2,400 | 2,000–2,600 | High (22–28%) |
| Central Asia (KZ, UZ) | 1,400–1,700 | 1,300–1,600 | Moderate (18–24%) |
| Southeast Asia (Thailand, Vietnam) | 1,400–1,700 | 1,000–1,400 | Low (12–18%) — high diffuse |
| Australia (inland) | 1,800–2,200 | 1,900–2,400 | High (22–27%) |
From GHI to plane-of-array (POA) irradiance
The yield modelling workflow translates GHI to the irradiance actually falling on your tilted panels — called POA (Plane of Array) irradiance. This requires three steps:
- Decomposition: split GHI into direct beam (Bh) and DHI using a decomposition model (Erbs, Reindl, or Perez). This is needed when only GHI is available and DNI/DHI are not separately measured.
- Transposition: convert beam and diffuse components from horizontal to the tilted surface angle using a transposition model (Hay-Davies, Perez, or Reindl). The Perez model is most accurate for diffuse transposition on tilted surfaces.
- Shading and soiling correction: apply near-shading horizon profiles and soiling losses — see our article on soiling loss factors — to arrive at net incident irradiance.
The result is hourly POA irradiance, which is then combined with module temperature coefficients and inverter efficiency curves to produce a P50 annual energy yield estimate.
P50 vs P90: understanding yield probability
A P50 yield estimate is the production level expected to be exceeded in 50% of years — the median. A P90 estimate (exceeded in 90% of years) accounts for interannual irradiance variability and is used as the conservative case in debt financing. The difference between P50 and P90 is typically 5–10% for well-characterised sites with long data records, widening to 10–20% in data-sparse regions where satellite model uncertainty is higher.
For our target markets in Central Asia and the Middle East — regions with high DNI, relatively low diffuse fraction, and rapidly expanding solar deployment — accurate GHI data at project sites is increasingly available from Solargis and PVGIS. Contact us to discuss irradiance data sourcing for your specific country and how we factor it into equipment sizing for commercial inverter and module selection.