The platform is under development. Coming soon!
Sentinel plants

Representative observation points for a distributed PV fleet.

SentinelPV reduces monitoring complexity by selecting a limited set of PV systems that can represent larger geographical areas.

The concept of a sentinel plant is the methodological core of the project. A sentinel is not simply a PV system located at the geometrical centre of an area: it is an observation point selected because its behaviour and context make it informative for a broader group of systems.

From many installations to meaningful spatial groups

The first step is to describe the PV fleet through multiple families of information. Historical production profiles show characteristic daily shapes, seasonal changes, variability and ramping events. Meteorological information describes irradiance, temperature, cloud conditions and local weather regimes. Geographical and topographical layers provide position, elevation, slope, aspect and other features that can influence local solar availability.

These variables are integrated into a multi-criteria spatial analysis. The project then compares alternative ways of grouping sites into coherent regions, including approaches that can preserve geographical continuity and relationships between neighbouring areas. The aim is not to create clusters for descriptive purposes alone, but to obtain regions that are useful for subsequent forecasting.

Selecting representative systems

Within each spatial group, the project identifies one or more monitored systems that can act as representative sites. A useful sentinel should reflect the dominant production behaviour of its zone while remaining based on real, measurable data. The selected sites become reference points for both model training and the generalisation of forecasts to systems that are not monitored directly.

Zones of influence

Each sentinel can be associated with a geographical zone of influence. These zones provide a direct connection between the statistical analysis and the GIS platform: users can visualise how the territory has been partitioned, which site represents each region and how the corresponding forecast contributes to a larger zonal estimate.

Finding the right spatial resolution

The optimal number of sentinel sites is not known in advance. Using more sentinels can preserve local differences and increase forecast detail, but it also requires more measurements, data transmission and computation. Using fewer sites simplifies the system but may hide important spatial variability. SentinelPV therefore evaluates forecast performance at different spatial resolutions to identify a practical balance between accuracy and monitoring effort.

In practice. The sentinel network is intended to transform a very large monitoring problem into a smaller and more informative network of representative observation points.