Home/About the project & partners

A cross-border answer to a cross-border wildfire.

Karst Firewall 5.0 is an innovative ecosystem-based plan to adapt to climate change in the Karst — promoting wildfire-resilient forestry supported by an Industry 5.0 approach, co-funded by the Interreg VI-A Italy–Slovenia Programme.

28
Months
6
Partners
€1.06M
Budget
€849K
ERDF
90%
Progress
2
Countries

15 April 2024 → 14 August 2026 · Policy Objective 2 (a greener Europe) · Specific Objective 4 — climate adaptation & disaster-risk prevention · Project ITA-SI0600146

The project

Wildfires & their governance.

The Karst Firewall 5.0 project addresses two linked challenges in the programme area: wildfires and their governance. It develops innovative action plans to preserve the health of the Karst region and maximise its resilience.

Wildfires here are primarily driven by ongoing climate change, and current firefighting strategies are becoming less effective. Because the study area straddles the Italy–Slovenia border, cross-border cooperation is essential — a need made urgent by the significant wildfire event of summer 2022. The project integrates socio-ecological and socio-technological systems, broadening dialogue among the responders, sectors and communities affected by fire.

OBJECTIVE 01

Develop an effective cross-border mechanism for wildfire prevention and management.

OBJECTIVE 02

Develop and use innovative digital systems — drones, satellite imagery, predictive AI and "electronic noses" — to monitor risk and support first responders.

OBJECTIVE 03

Promote cross-border adoption of these technologies for more effective prevention systems and management protocols, in line with Industry 5.0.

5.0
Industry · Society
Why “5.0”?

A 5.0 approach: technology beside people and nature

The “5.0” in Karst Firewall is deliberate. Where Industry 4.0 automates, Industry 5.0 and Society 5.0 put people and the planet back at the centre: technology that supports and empowers human work rather than replacing it. Here that means tools which quietly strengthen everyday wildfire prevention — sensors, AI and a digital twin woven together with the Karst landscape and the communities who care for it, in a way that stays simple, sustainable and human.

The partnership

Six partners, two countries

IUAV
Lead partner

Università IUAV di Venezia

Italy · project coordination

Project coordinator and scientific research. Project PM Massimiliano Granceri Bradaschia, PhD Arch. Pian.

Infordata
Project partner

Infordata Sistemi Srl SB

Italy · digital platform

Digitalisation and AI models — builds the platform and the fire-intelligence models.

ZRC SAZU
Project partner

ZRC SAZU

Slovenia · research (Research Centre of the Slovenian Academy of Sciences and Arts)

Scientific research and field operations with drones and Earth observation.

Duino Aurisina – Devin Nabrežina
Pilot municipality

Comune di Duino Aurisina – Občina Devin Nabrežina

Italy · cross-border municipality

Pilot site and use case — also procured equipment and ran on-the-ground wildfire interventions.

Miren-Kostanjevica
Pilot municipality

Občina Miren-Kostanjevica

Slovenia · municipality

Pilot site and use case — also procured equipment and ran on-the-ground wildfire interventions.

KID PiNA
Project partner

KID PiNA — Associazione Culturale ed Educativa

Slovenia · community & education

Community engagement, education and public outreach.

Associated partners

Inside the partnership

What each partner brings?

Six partners, each carrying a distinct part of the work — from the science and the platform to the budgets and the bricks on the ground.

Lead partner · coordinator & scientific lead

Università IUAV di Venezia

Beyond coordinating the consortium, IUAV led the project's core science: the cross-border climate-change assessment, the wildfire hazard and vulnerability analyses (with ZRC SAZU), the prevention abaco (action catalogue), and the participatory laboratories that tested it with the people who fight wildfires.

5 foundational studies authored or co-authored
Research & results →
Project partner · the digital backbone

Infordata Sistemi Srl SB

A Società Benefit that designed and built the whole platform and the wildfire-intelligence behind it: the Karst Fire Weather Index, a machine-learning ignition model that lifts wildfire detection from a 29% baseline to 77.5%, a daily risk map over ~40,000 grid points, and cross-border alerting by email, SMS and WhatsApp.

AUC 0.89 ignition model · 4 alert channels
How the KFWI works →
Project partner · science & Earth-observation engine

ZRC SAZU

The Research Centre of the Slovenian Academy of Sciences and Arts (ZRC SAZU) co-authored the cross-border climate-change assessment and the wildfire hazard and vulnerability analyses, built the cross-border remote-sensing burned-area database (171 historical fires from 30 years of satellite imagery), flew four seasons of multispectral drone surveys, and led the Slovenian pilot action in Miren-Kostanjevica.

171 burned-area maps · 4-season drone monitoring
Eyes on the Karst →
Pilot municipality · investments on the ground (IT)

Comune di Duino Aurisina – Občina Devin Nabrežina

Turned the plan into ground action after the 2022 wildfire: 9 ha of selective black pine clearing (~270 m³ of dead wood removed), procurement of electronic-nose sensors and drones that feed the platform, and reactivation of grazing to keep the Karst heathland open — while stewarding the 107-ha Falesie di Duino reserve.

9 ha restored · 107 ha reserve managed
Pilot sites →
Pilot municipality · investments on the ground (SI)

Občina Miren-Kostanjevica

Ground zero of the 2022 mega wildfire (~1,000 ha burned in the municipality alone), Miren-Kostanjevica co-financed three pilot actions — preventive grazing on the burn scar, dry-stone-wall restoration that doubles as firebreak and heritage, and dolina mosaic restoration — and has committed to scaling them past 2026.

3 pilot actions · 22-year land-use evidence base
Pilot sites →
Project partner · community & education

KID PiNA

PiNA designed and facilitated the participatory laboratories that brought about 60 farmers, foresters, firefighters and officials to the table, turning lived local knowledge into 29 municipal and 8 national prevention measures, and built gamified fire-awareness learning for young people.

~60 stakeholders engaged · 37 measures co-produced
Prevention →
The pilot sites

Where the plan meets the ground

Two municipalities — one in Italy, one in Slovenia, both scarred by the 2022 wildfire — put real budgets and real work behind the project's recommendations. This is what cross-border adaptation looks like on the ground.

~1,000 ha
burned in Miren-Kostanjevica alone in the 2022 cross-border mega wildfire.
9 ha
of post-wildfire forest under active restoration in Duino Aurisina.
3
pilot actions financed in Miren-Kostanjevica — grazing, walls, dolinas.
2
countries — one shared Karst, one shared method.
The July 2022 wildfire on the Karst, reaching the Italy–Slovenia motorway tollgate.
July 2022 — the cross-border Karst mega-fire reaches the motorway tollgate. Roughly 1,000 hectares burned across Italy and Slovenia; the recovery is what these two pilot sites are built on.
Pilot — Italy

Duino Aurisina – Devin Nabrežina: restoring a burned slope, sensor by sensor

On a 9 ha slope above the SS 55 — burned in 2022, now classed extremely vulnerable — the municipality is undoing decades of wildfire-prone planted black pines and bringing back the open Karst grassland.

9 ha of selective clearing

Crews removed wildfire-killed and high-risk black pine — about 270 m³ of dead wood — using low-impact rubber-tyred tractors and no earthworks, protecting the native oak and hornbeam already regenerating underneath.

Sensors & drones, procured locally

The municipality is acquiring electronic-nose (VOC / CO / CO₂) sensors and survey drones that feed straight into the Karst Firewall platform for early detection and recovery monitoring.

Grazing brings the heathland back

Targeted, subsidised pastoralism keeps woody encroachment down — a measure that reduces fuel and restores EU-protected dry-grassland habitat (6210) at the same time.

Stewarding 107 ha

Duino Aurisina also manages the Falesie di Duino reserve, home to Centaurea kartschiana — a plant that grows nowhere else on Earth.

Pilot — Slovenia

Miren-Kostanjevica: reading 22 years of land change — then reversing it

ZRC SAZU's analysis showed exactly why the 2022 wildfire spread so far: as farming was abandoned, the landscape filled with fuel. The municipality's three pilot actions target that directly.

−38%permanent meadows (2002–2024)
+297%trees & shrubs encroaching
104wildfires recorded 1998–2024
Preventive grazing on the burn scar

Sheep, goats and donkeys, with fencing and a water cistern, hold open the Nova vas burn area — backed by municipal budget and the national invasive-shrub decree.

Dry-stone walls as firebreaks

Volunteers trained in the UNESCO-listed craft are rebuilding walls — kept clear by a 10 m vegetation-free band — that slow surface wildfire and carry the Karst's cultural memory.

Doline mosaic restoration

Community clearing of an overgrown sinkhole (dolina) reopens a natural low-fuel patch and a biodiversity refuge.

Built to last past 2026

With ZRC SAZU, the municipality will report which measures worked, their upkeep cost, and how to fund a wider roll-out — municipal, national and EU.

Source: deliverables D2.2.1 (Duino Aurisina pilot actions, IUAV) and D2.2.2 (Miren-Kostanjevica pilot measures, ZRC SAZU & the Občina with PiNA), WP2, 2026.

The Karst Fire Weather Index (KFWI)

An index that actually sees the Karst.

Generic European wildfire-risk maps treat the Karst like anywhere else — and miss almost all of its fires. The KFWI pairs Karst-calibrated FWI severity with a machine-learning ignition model to fix that.

FWI predicts how severe a wildfire would be, but not where one starts. In the Karst, 52% of wildfires are human-caused, 97.3% ignite below Europe’s generic “High” FWI threshold, and 71% start with FWI below the median. That is why KFWI adds ignition probability to weather severity.

97.3%

of Karst wildfires occur below Europe's generic "High" wildfire-risk threshold — invisible to a one-size-fits-all index.

77.5%

wildfire detection with the KFWI, versus a 29% baseline using FWI alone.

0.89

AUC of the Random Forest ignition model behind the index.

Fire-weather severity

Copernicus-backed FWI forecasts establish daily wildfire-weather severity across the AOI.

ML ignition model

A Random Forest model estimates ignition probability from hotspot density, infrastructure distance and seasonally-corrected sampling.

Karst calibration

Severity and ignition are fused and locally calibrated to Karst fuels, terrain and climate.

Operational risk

The result drives the daily risk map, nowcasts and intervention planning — at street level.

See the KFWI on today's map → Open live KFWI statistics →
Early warning

An electronic nose that smells wildfire

Before a fire is hot enough for a thermal camera to see, it already smells. A cross-border network of low-cost electronic-nose sensors — mounted on trees in the most vulnerable zones — detects the volatile organic compounds (VOCs) of early combustion and raises the alarm in real time, often before any flame or smoke plume is visible.

A field electronic-nose (VOC) sensor node on the Karst

✦ Illustrative image generated with AI.

VOC

Smell, not heat

Each node continuously samples the air for the volatile organic compounds released by smouldering vegetation — the chemical fingerprint of a fire, minutes to hours before flames or a thermal signature appear.

ML

Trained on wildfire

An embedded machine-learning model is trained to tell the specific smell of a Karst wildfire from ordinary background air, so the network triggers a real-time alarm while keeping false positives low.

LoRaWAN

Autonomous in the field

Nodes communicate over long-range, low-power LoRaWAN and run on a small solar panel and battery — no mains power, no cabling. Each one also logs temperature, humidity, air pressure and VOC levels.

TWIN

Sharper simulations

Beyond the instant alarm, the live sensor stream feeds the digital twin: real micro-climate and early-detection data make the fire-spread simulator more precise.

Project coordination: Università IUAV di Venezia, Santa Croce 191, 30135 Venezia — Francesco Musco · karstfirewall@iuav.it. Digital platform: Infordata Sistemi Srl Società Benefit, Strada per Vienna 55/1, 34151 Trieste (TS), Italy.

Karst Firewall 5.0 — the cross-border wildfire digital twin — was designed and developed by Infordata Sistemi Srl Società Benefit. Its alert and monitoring features are offered free of charge to the public; access to the reserved area with the standard functions is also free for academia, researchers, students, public administration and NGOs. Advanced functions and dedicated support are available commercially, including for businesses — sales@infordata.it.

Licensing — Karst Firewall 5.0 follows an open-core model: the community platform and the project knowledge are openly licensed, while the advanced modules and the hosted service are commercial. Three layers, three licences:

Core community platform
EUPL-1.2-or-later — the platform source code is released as open source under a copyleft licence drafted by the European Commission and well suited to EU public projects.
Docs, methodology & datasets
CC BY 4.0 — the documentation, the KFWI methodology and the published datasets are free to share and adapt, with attribution.
Advanced modules & SaaS
Proprietary — the advanced / enterprise modules and the hosted (SaaS) service are offered under a separate commercial licence.