Objectives of the service
AI suitability surface across a target region in north-eastern Hungary. Image credit: GeoX Kft. / OPT4SOL.AI consortium.
OPT4SOL.AI develops and validates a decision-support service for the selection of locations for utility-scale photovoltaic (PV) plants. The service produces, for any chosen area of interest, a per-cell suitability score at 100 m resolution and a ranked shortlist of candidate sites, derived from Earth Observation data, national infrastructure and land-use layers, and a machine-learning model trained on existing PV installations. The Demonstration Project delivers the service end-to-end and validates it through a pilot with ENERGOSUN Investment Zrt across five Central and Eastern European countries: Hungary, Poland, Czechia, Croatia and Cyprus. The service is delivered as four product tiers from a single back-end: OPT4SOL.Viewer (free, anonymous web map of suitability layers); OPT4SOL.AI (registered web access, user-defined area of interest, point scoring); OPT4SOL.GIS (a QGIS plugin with the full data stack and spatial-analysis tools); and OPT4SOL.AI PRO (the GIS tier plus training of custom AI models on the user’s own labelled sites and proprietary data).
Users and their needs
OPT4SOL.AI is targeted at organisations involved in the planning and financing of utility-scale photovoltaic installations. The pilot user is ENERGOSUN Investment Zrt (Hungary). The wider target user community and their needs:
-
PV project developers and EPC contractors: ranking candidate sites across a national territory against grid feasibility, land quality, environmental constraints and demand proximity.
-
Investors and asset managers in renewable energy: repeatable methodology for pre-feasibility screening at national scale.
-
Consultants and GIS analysts supporting solar developers: harmonised, cross-border data layers and a workflow integrated with their existing QGIS environment.
-
Energy regulators and grid operators: transparent map publications to anchor public consultations on grid reinforcement and zoning.
The pilot covers five countries: Hungary, Poland, Czechia, Croatia and Cyprus.
Service/ system concept
The service produces, for any chosen area of interest, a per-cell PV suitability score at 100 m resolution and a ranked shortlist of candidate sites. Inputs include Copernicus Sentinel-1 and Sentinel-2 Earth Observation data and approximately 70 national and EU spatial data layers, covering high- and medium-voltage grid, substations, Natura 2000 and other protected areas, land cover, land quality, agricultural value, existing PV installations, terrain, population, demand proximity and similar. Training labels are sourced from public PV inventories (OpenStreetMap, the MEKH licensee list for Hungary, equivalents for the other countries).
The service is delivered through two client surfaces sharing one back-end. A web application provides the Viewer and AI tiers in any modern browser. A QGIS plugin provides the GIS and AI PRO tiers and adds project save, custom-data import, spatial analysis and — at the PRO tier — training of custom AI models. The back-end is deployed cloud-agnostically using containers, managed PostgreSQL/PostGIS, S3-compatible object storage and a Celery/Redis job queue. The high-level system architecture is shown below.
Space Added Value
OPT4SOL.AI uses Copernicus Sentinel Earth Observation data. The satellite data serves three roles: it characterises land use, land cover and seasonality at fine spatial resolution across an entire country, at an update frequency that register-based data sources do not match; it provides input features to the AI suitability model, alongside infrastructural, grid and land-quality data; and it anchors the training labels. The combined use of Sentinel data and national reference layers across five countries produces a cross-border-harmonised view of PV suitability that no single national dataset offers.
Current Status
Sentinel-2 bands and derived spectral indices feeding the model for a candidate site.
The project is in the Baseline Design Review (BDR) phase, following kick-off in March 2026. A pilot agreement is in place user-needs and pilot-scoping discussions are in progress. The Requirements Document and System and Service Architecture have been prepared at developer-implementable depth and capture 64 system requirements across 12 areas. An architecture pivot has been completed: a custom web-based developer GUI has been replaced with a QGIS plugin plus a web application, reducing four product tiers to two implementation surfaces sharing one back-end. Work in progress includes finalising the per-country data-source lists for Poland, Czechia, Croatia and Cyprus, closing the open technical decision points ahead of the Critical Design Review (CDR), and producing the first UI prototypes.
Prime Contractor(s)
