Solar irradiance explained: GHI, DNI and why they matter

Every solar project begins with a question about the sun: how much energy reaches this site over a typical year? The answer is expressed through a handful of irradiance measures that appear in every feasibility study. Understanding them helps landowners, developers and investors read those studies with confidence.

Irradiance and irradiation

Two closely related terms are often used interchangeably:

• Irradiance is the instantaneous power of sunlight per unit area, measured in watts per square metre (W/m²).
• Irradiation (or insolation) is irradiance accumulated over time, usually expressed in kilowatt-hours per square metre (kWh/m²) per day or per year.

Feasibility studies usually quote annual irradiation, because energy yield over a year is what matters commercially.

The three components

Sunlight reaching the ground arrives in two ways — directly from the sun's disc, and scattered by the atmosphere and clouds.

The three are related: GHI is the diffuse component plus the direct component projected onto a horizontal surface.

Plane-of-array irradiance

Modules are rarely horizontal. They are tilted, and on tracking systems they follow the sun. The irradiance actually striking the module surface is called plane-of-array (POA) irradiance. Energy yield models convert GHI, DNI and DHI into POA irradiance for the chosen tilt or tracking strategy — which is why trackers versus fixed-tilt is partly an irradiance question.

Where the data comes from

• Satellite-derived datasets cover long periods and large areas, and are the usual basis for early feasibility work.
• Ground measurements at or near a site, using pyranometers, help validate and calibrate satellite data.
• Public reference datasets are available from national and international agencies.

For bankable studies, developers typically combine long-term satellite data with on-site or nearby ground measurements to reduce uncertainty.

From irradiance to energy yield

Irradiance is only the first step. An energy yield model then accounts for:

• Module orientation, tilt and tracking
• Shading from terrain, obstructions and adjacent rows
• Temperature effects on module efficiency
• Soiling losses from dust (module cleaning)
• Inverter and cable losses
• Module degradation over time (degradation and warranties)
• Plant availability

The result is usually expressed as expected annual energy at different confidence levels — a central estimate and more conservative figures that lenders use.

What this means for site selection

Regional irradiance in much of southern India is generally favourable, so the differences between candidate sites often come down to factors other than the sun: grid distance, shading, soiling, temperature and land conditions. That is why our land suitability checklist treats irradiance as one input among several, rather than the deciding one.

Uncertainty: P50, P90 and why lenders care

No estimate of future sunshine is exact. Year-to-year weather varies, and every dataset and model has some error. Energy yield studies therefore express output as a probability distribution:

• P50 — the central estimate: in any year, output is equally likely to be above or below it.
• P90 — a more conservative figure that output is expected to exceed in nine years out of ten (or over the long term, depending on how it is defined).

Lenders usually size debt on conservative figures such as P90, because they need confidence that the project can service its loans in poorer years. Reducing uncertainty — through better data, longer records and on-site measurement — can therefore improve how a project is financed, not only how accurately it is forecast.

Seasonal patterns

In much of southern India, solar generation follows a clear seasonal rhythm: strong output in the dry, clear months, and lower output during the monsoon when cloud cover increases and diffuse light makes up a larger share of the total. Seasonal variation matters for:

• Cash flow — revenue is uneven through the year
• O&M planning — cleaning needs are highest in the dry, dusty season
• Industrial consumers — on-site solar covers less of the demand during the monsoon, which affects how much grid power is still needed

Common mistakes in reading resource data

• Comparing figures from different datasets as if they were equivalent
• Treating a single year of measured data as representative of the long term
• Ignoring local shading, which no regional dataset captures
• Using GHI for a tracking plant without converting to plane-of-array irradiance
• Forgetting that hotter sites lose more output to temperature effects

How we use it

When we source and qualify land, irradiance data is part of the early technical screen, alongside shading, terrain and grid access. Detailed resource assessment and yield modelling are carried out by qualified technical partners as a project moves towards investment.