DeepCube · Project

AI Platform That Predicts Wildfires, Droughts, and Infrastructure Risks from Satellite Data

environmentTestedTRL 5

deepcube-h2020.eu

Imagine having a weather app, but instead of telling you it will rain tomorrow, it warns you weeks ahead that a wildfire or drought is coming — and explains exactly why. That's what DeepCube built: an AI system that chews through massive amounts of European satellite data (Copernicus) and spots dangerous patterns before they become disasters. The clever part is that it doesn't just give predictions — it shows its reasoning, so you can actually trust the results. They tested it on real problems like Mediterranean wildfires, African droughts, volcanic activity, and even infrastructure damage detection.

By the numbers

Use cases tested (drought, migration, fire, volcanic, tourism)

Consortium partners across 6 countries

SMEs in the consortium

Total project deliverables produced

56%

Industry partner ratio in consortium

The business problem

What needed solving

Businesses exposed to climate risks — insurers, infrastructure operators, tourism boards — currently rely on historical data and reactive assessments to manage wildfire, drought, and ground instability threats. By the time traditional monitoring spots a problem, the damage is already underway and costs are locked in. Companies need predictive tools that flag risks early and explain their reasoning clearly enough to justify business decisions.

The solution

What was built

An open, cloud-deployable AI platform (delivered in two versions with full documentation) that processes massive Copernicus satellite data through explainable deep learning pipelines. Includes trained models for fire risk forecasting, drought prediction, volcanic deformation detection, infrastructure monitoring via SAR data, and sustainable tourism assessment — plus dedicated training datasets, data cubes, and ontologies.

Audience

Who needs this

Property and catastrophe insurers covering wildfire and drought zonesCritical infrastructure operators (pipelines, dams, bridges) needing ground deformation monitoringRegional tourism authorities in climate-sensitive Mediterranean areasEnvironmental consulting firms serving industrial clients with climate risk assessmentsCivil protection agencies responsible for natural disaster early warning

Business applications

Who can put this to work

Insurance & Reinsurance

enterprise

Target: Property and agricultural insurers covering Mediterranean or African regions

If you are an insurer dealing with mounting wildfire and drought claims across Southern Europe and Africa — this project developed AI pipelines for short-term fire hazard forecasting in the Mediterranean and extreme drought prediction in Africa. The explainable AI component means your underwriters can see exactly why a region is flagged as high-risk, making it defensible for pricing decisions. The system was tested across 5 use cases with real Copernicus satellite data.

Infrastructure & Utilities

mid-size

Target: Pipeline operators, dam managers, and critical infrastructure owners

If you are an infrastructure operator worried about ground movement damaging pipelines, bridges, or buildings — this project built automated deformation detection using satellite radar data (interferometric SAR). It can spot millimeter-level ground shifts and trend changes on critical infrastructure over time. The platform was delivered in two versions with full documentation, ready for cloud or HPC deployment.

Tourism & Hospitality

any

Target: Regional tourism boards and resort management companies in climate-sensitive areas

If you are a tourism authority struggling to plan around increasing climate disruptions like heatwaves, droughts, or wildfires — this project developed a dedicated use case for sustainable and environmentally-friendly tourism using Copernicus services. The platform combines satellite data with AI to help you assess environmental risks to tourism assets and plan seasons around real climate data rather than guesswork.

Frequently asked

Quick answers

What would it cost to use this platform?

The DeepCube platform was developed as an open and interoperable system under EU public funding (RIA scheme). Based on available project data, the platform can be deployed on existing cloud infrastructure or HPC environments including DIAS, which suggests operational costs would primarily be compute and data access fees rather than licensing.

Can this scale to cover my entire portfolio or region?

The platform was designed to handle big Copernicus data — the full European satellite observation archive. It was tested across 5 use cases spanning Africa and the Mediterranean, covering drought, fire, volcanic activity, infrastructure monitoring, and tourism. The architecture supports distributed deep learning via the Hopsworks platform, so it is built for scale.

What about IP and licensing?

DeepCube was funded as an RIA (Research and Innovation Action), which typically means results are openly accessible. The consortium of 9 partners across 6 countries includes 4 SMEs and 5 industry partners, so commercial licensing arrangements may exist for specific components. Contact the coordinator for specific terms.

How do I know the AI predictions are reliable?

A key differentiator is the Explainable AI and Causality component — the system doesn't just predict, it shows why. The project also developed hybrid models that respect physical laws, meaning the AI can't produce physically impossible results. This makes outputs auditable and defensible for regulatory or business decisions.

How long would integration take?

The DeepCube Platform was delivered in two versions (v1 and v2) with user manual and full documentation. It is designed to be interoperable and deployable on several cloud infrastructures and HPC environments including DIAS. Based on available project data, integration depends on your existing data infrastructure.

Is there ongoing support or development?

The project closed in December 2023. The consortium included 9 partners with 56% industry ratio, and 4 SMEs were involved — suggesting commercial interest in continuing the work. Based on available project data, check the project website at deepcube-h2020.eu for post-project developments.

What data do I need to provide?

The system is built on Copernicus satellite data, which is freely available from the EU. The platform includes its own training datasets, data cubes, and ontologies (delivered in two versions). For infrastructure monitoring, you would need to specify locations of interest for the SAR deformation analysis.

Consortium

Who built it

The DeepCube consortium brings together 9 partners from 6 countries (Germany, Greece, Spain, France, Italy, Sweden), with a strong industry presence at 56% — 5 industry partners including 4 SMEs. The coordinator is the National Observatory of Athens, a well-established Greek research institution. With 2 universities and 2 research organizations providing the scientific backbone, and the majority being industry players, this consortium is positioned to move results toward commercial use. The geographic spread across Southern and Northern Europe mirrors the project's focus areas (Mediterranean fire risk, African drought monitoring), giving it practical testing grounds across climate zones.

ETHNIKO ASTEROSKOPEIO ATHINONCoordinator · EL
MURMURATIONparticipant · FR
ETHNIKO KAI KAPODISTRIAKO PANEPISTIMIO ATHINONparticipant · EL
HOPSWORKS ABparticipant · SE
UNIVERSITAT DE VALENCIAparticipant · ES
INFALIA PRIVATE COMPANYparticipant · EL
MAX-PLANCK-GESELLSCHAFT ZUR FORDERUNG DER WISSENSCHAFTEN EVparticipant · DE

How to reach the team

Coordinator is the National Observatory of Athens (Greece). Use SciTransfer's lookup service to find the right contact person.

Order a report on this consortium See ETHNIKO ASTEROSKOPEIO ATHINON profile →

Next steps

Talk to the team behind this work.

Want to explore how DeepCube's AI-driven satellite analytics could reduce your climate risk exposure? SciTransfer can connect you directly with the team — contact us for a briefing.

Order a report on this topic More in Environment & Climate