Research and EU projects

SEBELO is built through applied research with forestry companies, researchers and fire services across Europe. These are the projects behind the sensors and the platform.

155.8 m
average error in locating where a fire started, in JP20 field tests
10
controlled smoke tests at five pilot sites in Latvia
14
partners from 13 EU regions in the IT4EST project
5
countries with live IT4EST demonstrations of SEBELO

JP20: finding where a forest fire starts

Development of an intuitive territory monitoring platform with AI-based sensor integration for forestry and agriculture

JP20 builds a prototype platform that uses artificial intelligence to give early warning of the main environmental risks in forestry and agriculture. It works with different LoRaWAN sensors and with outside data such as weather data and maps. SEBELO's focus in the project is the most critical of these risks: finding where a forest fire started while it is still small.

Framework: Research project of the Forest Sector Competence Centre (Latvia)
Funding: EU Recovery and Resilience Facility, Recovery Fund project No. 5.1.1.2.i.0/2/24/A/CFLA/001
Partners: SIA Falkors Group (SEBELO), SIA ZLM, SIA Animas
Period: October 2025 – October 2026

How the point of origin is estimated

Sensor
More smoke measured
Wind
Real origin
Estimated origin

Illustration. Sensors downwind of the fire measure more smoke. Combining their readings with the wind direction narrows down where the smoke came from.

Activity 1: industrial pre-study

January–March 2026Completed

Surveyed 28 respondents in Latvia: farmers, forest owners and municipalities.
Consulted civil protection services and regional authorities in Portugal, France, Spain and Greece.
Assessed more than 60 LoRaWAN sensor models in six risk categories and 15 external data sources.
Ran field tests and collected a dataset of more than 5,500 readings.
Defined the platform's functional requirements and the data the AI models need.

Finding: in Latvia, pests, drought and flooding are the main risks; in Southern Europe, wildfires are the most critical. Early detection of where a forest fire starts was chosen as the priority for the AI work.

Activity 2: AI analytics for fire origin

April–October 2026In progress

An AI tool combines the readings of the LoRaWAN sensor network with wind data and estimates in real time where the fire started. It uses a Bayesian geospatial model with a Gaussian plume likelihood.
Seven software modules, including data quality control, anomaly detection, wind data from the Latvian Environment, Geology and Meteorology Centre (LVĢMC) and calibration.
Ten controlled field tests at five pilot sites in Latvia with ZLM: a smoke source was lit at GPS coordinates recorded in advance, and the tool's estimate was compared with the real point.
Explainable AI: every estimate comes with a short text explanation based on the sensor readings and how they line up with the wind direction.
Animas is preparing the platform prototype: technical architecture, API specification, and an isolated connection so the AI module cannot disturb the existing platform.

Milestone: the prototype has reached technology readiness level 4 (TRL 4), validated in the field.

Field test results

155.8 m
average distance between the estimated and the real point of origin
90%
of tests within the 200 m target, better than the project application required
329 → 156 m
average error before and after calibration

Next steps

Add EU-DEM terrain data, so slopes and valleys are taken into account.
Further field tests in different forest types and weather.
Build the platform prototype around the AI module.

IT4EST: European test beds for forest technology

Interregional Tech for forEst SusTainability

IT4EST helps European regions put new forest technologies into real use: preventing, limiting and repairing damage from wildfires and insect pests. The partners combine early warning, artificial intelligence, robotics, remote sensing and firefighting technology and test them together on test beds across Europe.

Programme: Interregional Innovation Investments (I3) instrument, co-funded by the European Union through the European Regional Development Fund
Grant agreement: No. 101160993
Consortium: 14 partners from 13 EU regions, eight of them less developed regions
SEBELO's participant: SIA Falkors Group, Latvia
Period: September 2024 – August 2027

Project website

IT4EST project partners at the kick-off meeting in September 2024 — The project started with the partners' kick-off meeting in September 2024. About the kick-off

SEBELO's part

SEBELO brings the ground sensor network for early wildfire detection and shows it working live in different landscapes and climates, together with the other partners' technologies.

Live demonstrations

Romania2024
First public demonstration of the SEBELO system within IT4EST.
Read the report
PortugalMarch 2026
Field test at the Centro Ciência Viva da Floresta.
Read the report
SpainMay 2026
Live demonstration at RESIST in Valencia.
Read the report
Greece2026
IT4EST Attica test bed at the Dionyssos marble quarries: 12 sensors in marble dust and strong wind, alongside fire suppression and incident coordination technologies.
Read the EU article
FranceSeptember 2026
Live test at the Meymac Test Bed.
Read the report

Where it started

SEBELO's first research project, "Development and operational research of LoRa-based air quality sensor node for forest fire monitoring" (project No. VP-PI-2022/5), was carried out by SIA Sebelo under agreement No. SKV-L-2022/189 of 10 May 2022 with the Investment and Development Agency of Latvia, within the measure "Promotion of International Competitiveness", co-financed by the European Regional Development Fund.

A research partner or pilot site?

We work with forest owners, municipalities, fire services and research institutions on pilots and joint projects. Tell us about yours.