The project was developed by the CIN Researcher Antonio Pascarella

Project Description

In 2023, global photovoltaic (PV) installed capacity—a key driver in the decarbonization of the power sector—reached 1,552.3 GWp. In France, the capacity stood at 19.9 GWp as of April 2024. Over the previous year, PV capacity grew by nearly 32% worldwide and by 15.7% in France. Despite this rapid expansion, the integration of PV electricity into power grids remains challenging, primarily due to limited knowledge about rooftop PV installations. These systems account for around 20% of France’s total PV capacity, yet their power production is rarely measured directly, leading to what is known as a lack of observability.

In this study, we address this challenge by leveraging high-resolution ground-truth measurements from individual rooftop PV systems—available at unprecedented temporal and spatial scales. We demonstrate that it is possible to estimate the power production of such systems accurately by combining solar irradiance and temperature data with basic system characteristics inferred from remote sensing, alongside a simplified irradiance-to-electricity conversion model. Our approach achieves a normalized estimation error (pRMSE) of approximately 10% relative to system capacity.

These findings highlight that even with minimal information and a simple modeling framework, it is possible to significantly improve the observability of rooftop PV systems. Enhancing this observability is essential for better integration of distributed solar power into the electric grid. More broadly, our results suggest that reliable estimates of small-scale PV generation can be obtained without detailed system-level data.

In addition, during the project, some work was done on onboard AI methods for drought detection. A second activity involved the drafting of a scientific review (which is still ongoing) for Earth Observation applications for energy and the environment.