{"id":157882,"date":"2025-07-28T10:48:07","date_gmt":"2025-07-28T10:48:07","guid":{"rendered":"https:\/\/www.pv-tech.org\/?p=157882"},"modified":"2025-07-28T10:48:09","modified_gmt":"2025-07-28T10:48:09","slug":"from-assumptions-to-accuracy-pv-industry-needs-physics-based-solar-data","status":"publish","type":"post","link":"https:\/\/www.pv-tech.org\/from-assumptions-to-accuracy-pv-industry-needs-physics-based-solar-data\/","title":{"rendered":"From assumptions to accuracy: Why the PV industry needs physics-based solar data"},"content":{"rendered":"\n
\"A
The use of better-quality data can improve the business case for a number of solar projects. Image: Photon Group.<\/figcaption><\/figure>\n\n\n\n

The integrity of a PV project largely depends on the quality of the solar, meteorological and environmental data used. Whether modeling PV performance, optimising system design or assessing financial risk, the foundation of decision-making rests on one crucial question: Are we using the best available science to characterise real-world solar resource conditions, or are we continuing to rely on legacy approaches and black-box assumptions?<\/p>\n\n\n\n

As much as it pains me to admit, many players in the solar energy industry still work with empirical, low-quality solar datasets, built on various simplifications, often with crucial details being opaque or unvalidated. In energy modelling, a patchwork of solar datasets is used \u2013 some are so outdated, they belong in a museum.<\/p>\n\n\n\n

Even worse, there are some in the industry who cherry-pick data that best supports their business case, then invent \u2019expert\u2019 justifications to defend those choices.<\/p>\n\n\n\n

However, as solar energy becomes a mainstream source in markets that demand accuracy, transparency and accountability, this approach is not only wrong, it is also dangerous. Solar developers must move decisively toward scientifically rigorous solar model data if it is to deliver on its promise of sustainability and resilience.<\/p>\n\n\n\n

<\/a>Why empirical solar modeling and ad-hoc approaches fall short<\/h2>\n\n\n\n

Simple empirical models, by definition, depend on statistical correlations, not physical causality. They may work in familiar and simple conditions, for which they were initially developed, but they break down when applied to new geographies. Some consultants routinely mix and match solar data from different sources and convert them into synthetically generated hourly data.<\/p>\n\n\n\n

These approaches are subjective and result in unrealistic, misleading data products that fail in complex environments with high solar variability. This is not physics-based modeling; it\u2019s guesswork<\/a>.<\/p>\n\n\n\n

This practice is neither traceable nor reproducible \u2013 key attributes of scientific approaches. When physical relationships are violated, such as the fundamental balance between global, direct and diffuse irradiance, it\u2019s like a three-legged stool with mismatched legs: it somehow stands, but not straight, nor is it safe to sit on.<\/p>\n\n\n\n

<\/a>The case for physics-based solar modeling<\/h2>\n\n\n\n

Physics does not bend to business cases \u2013 and that is its greatest strength. It reveals the truth.<\/p>\n\n\n\n

When solar irradiance is modeled using scientific principles \u2013 following the radiative laws in the atmosphere; accounting for attenuation by aerosols and water vapor; and incorporating cloud dynamics and high-resolution terrain data \u2013 the results are not only physically grounded but also reproducible and independent.<\/p>\n\n\n\n

Modern solar models are built upon rigorous scientific equations, drawing on data from meteorological satellites, global weather models and other geospatial datasets. These models are designed to perform accurately across all climate zones. Machine learning has emerged as a next-level approach in solar modelling, particularly where physical models and globally available input data begin to reach their limitations.<\/p>\n\n\n\n

\"A
The highest-quality solar, meteorological and environmental data come from validated physics-based solar models that deliver consistent accuracy across all climates and geographies. Image: Solargis.<\/figcaption><\/figure>\n\n\n\n

The rigorous physics-based models do not require further manipulation to produce acceptable results. Instead, they can be validated against high-quality ground measurements and, for local conditions, refined through site-specific model adaptation, rather than speculative tuning. This level of scientific rigor<\/a> is increasingly demanded by financial stakeholders, a welcome development that strengthens the integrity of the solar industry.<\/p>\n\n\n\n

Lenders and investors are no longer satisfied with generic models. They now require objective, physics-based simulations that are transparent and can be independently validated. Many even request formal \u2018reliance letters\u2019 as evidence that the data and modeling practices behind a project\u2019s financial case are science-driven and free from commercial bias.<\/p>\n\n\n\n

This shift is helping to eliminate providers that offer unverified or inconsistent data, improving overall credibility across the market<\/a>.<\/p>\n\n\n\n

<\/a>From Linke Turbidity to new generation solar models<\/h2>\n\n\n\n

About 15 years ago, the Linke Turbidity Factor was commonly used to estimate how aerosols in the atmosphere reduce the amount of sunlight reaching the ground. This atmospheric parameter gives a rough idea of how \u2019polluted\u2019 the atmosphere is by comparing it to a number of clean, dry atmospheres that would cause the same dimming of sunlight.<\/p>\n\n\n\n

But in practice, it was hard to get accurate values because clear-sky measurements were rare, and the data was often limited to monthly averages. These models also had low geographic availability and, because of poor resolution, couldn\u2019t account for local conditions, making them too simplistic for accurate solar resource assessment.<\/p>\n\n\n\n

Today, modern physics-based models go much further. The new generation of solar models simulates the transmission of solar radiation through the atmosphere using detailed atmospheric data and complex cloud interactions sourced from global weather models and satellite observation systems.<\/p>\n\n\n\n

This enables accurate solar irradiance estimates even in highly complex regions, such as West Africa, Indonesia, Northern India, or the Amazon Basin \u2013 areas where empirical models have historically failed due to high short-term weather variability and critical weather events. It is important to note that the experience gained through years of working on solar projects worldwide has greatly contributed to the development of more reliable solar data.<\/p>\n\n\n\n

Crucially, the modern solar radiation models are sensitive. They can detect local anomalies like wildfire smoke or industrial haze, but that sensitivity also demands care: the more advanced the model, the more it relies on high-quality input data and expert calibration. Think of them like high-performance engines \u2013 they deliver more, but only if properly controlled and maintained.<\/p>\n\n\n\n

<\/a>What should solar developers and investors demand?<\/h2>\n\n\n\n

To design bankable, high-performing PV projects, stakeholders must be more selective when choosing solar data providers. This means:<\/p>\n\n\n\n