EO x Grid | ESA Space Solutions

Objectives of the service

Electricity grid operators face mounting challenges in managing distributed solar energy, particularly due to limited visibility into behind-the-meter photovoltaic (PV) production and the lack of accurate short-term forecasts. These gaps hinder effective planning, real-time monitoring, and coordination with national transmission operators. Our service addresses these issues by delivering high-resolution PV generation forecasts and reconstructing true consumption profiles using satellite data, artificial intelligence models, and minimal reference sensors.

The solution integrates Earth observation imagery, numerical weather predictions, and grid metadata to produce feeder-level forecasts and load reconstructions without requiring costly hardware upgrades. It enables grid operators to optimize infrastructure investments, reduce imbalance penalties, and improve hosting capacity for renewable energy.

The activity is a feasibility study aimed at validating the technical and operational impact of these services through two use cases with a regional grid operator. It includes model development, stakeholder engagement, and business impact assessment, with the goal of preparing for scalable deployment across European electricity networks.

Both technical and commercial viability of the envisaged service have been confirmed in the Feasibility Study.

Major Achievements of the Feasibility study

Technical Superiority: The core technical achievement is the 61% reduction in Weighted Root Mean Squared Error (wRMSE) for PV forecasting (UC2) achieved by the inclusion of satellite-derived features compared to the simple baseline model. This proves that the EO fusion approach delivers operationally useful accuracy for real-time grid management.
Confirmed Market Pull: Securing a Letter of Interest (LoI) and confirmed Willingness to Pay from the pilot partner is a decisive major achievement. This commitment de-risks the commercial roll-out and provides a powerful reference case.
Quantified ROI: The project successfully quantified a clear, compelling Return on Investment (ROI) for the customer.

Users and their needs

The primary user community targeted by the service is electricity distribution system operators (DSOs), with Elektro Primorska in Slovenia serving as the representative user throughout the feasibility study. These operators manage regional electricity grids and face increasing challenges due to the rapid growth of distributed photovoltaic (PV) systems. Target markets are European Union markets.

User involved at DSOs:

Grid planning engineers
Operational analysts
Strategic development teams at DSOs

User Needs:

Accurate reconstruction of gross electricity consumption hidden by behind-the-meter PV production
Reliable short-term PV generation forecasts for unmonitored assets
Improved visibility of distributed energy resources without costly hardware upgrades
Enhanced data inputs for planning, state estimation, and coordination with national transmission operators

Project Challenges:

Integrating diverse datasets (satellite, weather, grid metadata) with limited historical measurements
Achieving high forecast accuracy across varied PV asset types and locations
Ensuring compatibility with existing grid management systems
Demonstrating regulatory and operational value to justify adoption

Service/ system concept

The EO x Grid service provides electricity grid operators with two key capabilities: (1) reconstruction of true electricity consumption hidden by rooftop solar systems, and (2) accurate short-term forecasts of photovoltaic (PV) generation across the grid. Users receive clean, time-stamped data streams showing gross load per customer and predicted solar output per asset, enabling better planning, real-time monitoring, and coordination with national transmission operators.

How it works:

The system collects satellite images, weather forecasts, and grid data. It uses artificial intelligence to estimate how much solar energy is being produced and adds that back to smart meter readings to reveal actual electricity use. It also predicts future solar output using updated weather data and reference sensors.

High-Level Architecture:

Data Inputs: Satellite imagery, weather forecasts, smart meter data, PV metadata
Processing Engine: Cleans and aligns data, runs forecasting models
Outputs: Load reconstruction and PV forecasts delivered via dashboards or APIs

Here’s a simplified diagram:
This modular setup ensures scalability, accuracy, and minimal hardware requirements.

Space Added Value

The EO x Grid service uses Earth observation data from two key space assets: EUMETSAT’s Meteosat satellites (MSG/MTG series) and the Copernicus Climate Data Store (CDS). These satellites provide high-frequency, high-resolution imagery and atmospheric variables such as cloud cover, solar radiation, and temperature—critical inputs for estimating and forecasting photovoltaic (PV) generation.
Added Value of Space Assets:

Temporal granularity: Meteosat delivers hourly updates, capturing rapid changes in irradiance that ground-based sensors or weather models often miss.
Spatial coverage: Satellite data ensures consistent visibility across all grid areas, including remote or unmonitored PV installations.
Cost-efficiency: Unlike hardware-heavy solutions, satellite data enables scalable forecasting without installing new sensors.

Why It Outperforms Competitors:

Most existing providers rely solely on weather models or limited ground sensors, which lack the precision and coverage needed for feeder-level forecasts.
EO x Grid combines satellite imagery with minimal reference sensors and AI models to reconstruct true consumption and predict PV output—delivering superior accuracy and operational relevance for grid operators.

This integration of space assets enables a meterless, scalable, and regulation-aligned solution for managing distributed solar energy.

Current Status

Between April 2025 and January 2026, the EO x Grid feasibility study completed seven development sprints and eight stakeholder workshops, including four co-design sessions with Elektro Primorska (DSO).

Two use cases were finalized—gross load reconstruction (UC1) and short-term PV forecasting (UC2)—and validated through model development and business impact analysis. Over 260 satellite variables and 24 weather features were processed, with the best-performing models reducing forecast error by over one third compared to baseline model for UC1 and by over two thirds compared to baseline model for UC2. A dedicated server and database were deployed to support scalable forecasting.

The strong collaboration with Elektro Primorska (DSO) ensured that the service concept was not just theoretically possible and technically feasible but was co-developed to align seamlessly with the DSO’s real-life problems, existing operational ecosystem and critical financial Key Performance Indicators (KPIs). Applying UC1 restores visibility of actual demand, enabling DSOs to optimize and defer 1–3% of annual network reinforcement investments. Applying UC2 enhances operational tools (e.g., state estimation) and is expected to reduce LV grid technical losses by 4–6% annually.

We have developed a business model for the planned service, together with the business plan for the coming years. The business plan envisages a demonstration phase with clear milestones and deliverables as a means of developing fully functioning solution. The deployment will be supported by an experienced and strong team.

The results of the Feasibility Study confirmed both commercial as well as technical feasibility of the envisaged service.