This blog article provides an overall description of the fourth and final SAFERS demonstration pilot, which took place in Thessaloniki (Greece), including key preparation aspects, the dynamics during the execution, as well as an analysis of the feedback collected from the stakeholders that attended the event.
The demo was conducted on December 12, 2023, through an exercise that was designed for testing and providing feedback on the second release of the β2 version of the SAFERS integrated system. The exercise consisted of wildfire risk scenario storyline that was inspired by the fire in Seich Sou Forest of Thessaloniki in 1997. Due to lack of data from that period, data from other more recent forest fires in other areas in Greece were used to simulate and demonstrate the platform’s functionalities.
In total, 16 organizations (13 from Greece, 1 from Cyprus, 1 from Bulgaria and 1 from North Macedonia), most of them directly involved in wildfire risk management operations, attended the event, mainly as observers.
1. Preparation aspects
The overall coordination for the Thessaloniki demonstration was carried out by Hellenic Rescue Team (HRT) which is a volunteer non-profit Search and Rescue organization dedicated in helping people in distress. It was founded in 1994, with a potential manpower of 2000 trained members in various rescue specialties, making it the largest domestic volunteer organization. Its headquarters are based in Thessaloniki and has 34 branches all over Greece.
The event required a venue that could accommodate approximately 100 people and could also act as a C3 center. The new hall used by the Regional Council of Central Macedonia was selected. The hall is fully equipped with audio and speaker systems and offers the facilities to support simultaneous translation. These attributes have improved the flow of the event and allowed easier communication between local stakeholders and technical partners.
Region of Central Macedonia council hall (event’s venue)
Regarding the fire simulation, the selected location had to respect several characteristics, namely its proximity to the burnt area of 1997 and its visibility from the two cameras to allow the system to trigger an alert and be able to test both cameras.
Location of cameras (in blue) and fire simulation (in red)
For the demo’s needs, training sessions for HRT & HMOD staff as well as citizens, that would participate in various tasks, had to be organized:
- HRT staff that would operate the dashboard in C3 room during the demo had to be trained and for that, four training sessions took place in November 2023.
- The HRT & HMOD staff that would act as first responders had to be trained for the use of Chatbot as professionals during the demo.
- 70 participants had to be trained for the use of Chatbot as citizens.
2. The dynamics during the execution
The final demonstration of SAFERS project in Thessaloniki was attended by 75 people. Among them, 41 were from the SAFERS consortium, while the remaining 34 represented 16 organizations from Greece, Cyprus, Bulgaria, and North Macedonia mainly involved in wildfire management operations.
Involved organizations and projects
Moreover, to support, facilitate, and execute the demo storyline, the following assets from the hosting organizations (HRT and HMOD) were used: 7 members from HRT (that acted as first responders), 5 members from HMOD (that acted as first responders), 2 vehicles from HRT and 1 vehicle from HMOD.
A wildfire risk scenario storyline was designed bringing together managerial and operational procedures across all the phases of the Disaster Risk Management (DRM) cycle: Prevention and Preparedness, Detection and Response, and Restoration and Adaptation. As a result, the storyline was arranged in 11 sequential scenes, beginning with the synoptic risk scenario, followed by the outbreak of the fire and the response operations undertaken by the emergency groups, and concluding with the post-fire monitoring assessment in the aftermath of the incident. Every scene utilised at least one of the SAFERS Intelligent Services (IS), illustrating how they could be used in decision-making processes connected to wildfire risk management situations.
2.1. Exercise scenes I: Prevention and Preparedness
The scene takes place before the fire occurs. A meteorological scenario with typical weather conditions in the region that can lead to a potential wildfire risk scenario was simulated. This accentuated the necessity for heightened situational awareness and enhanced resource planning. The SAFERS IS and tools used were the weather forecast service to consult the medium and long-term (i.e., sub-seasonal) weather forecast maps.
The semantic reasoning service was also employed to support decision makers to take preparatory actions in case the weather conditions set a high fire risk as it was defined by SAFERS dashboard system. The IS automatically generated alarm messages which users examined to respond appropriately. The Fire Weather Index (FWI) employed in this situation was the one used by the Greek Civil Protection authorities to develop and broadcast alarm signals directed to the public and first responders separately.
The data generated were primarily pre-defined warnings for various fire danger levels. Messages directed to citizens were instructional to advise them on preventive and protective measures. Messages targeted to first responders were operational directives to carry out preventive and surveillance or patrolling activities in preparation for a possible fire incident.
2.2. Exercise scenes II: Detection and Response
The goal of the scene was to test and validate the fire detection cameras that were installed. To that end, fire simulation was realized in a pre-selected location. SAFERS ISs were involved through the two AI detection cameras that were installed in locations that overlook a part of the Chortiatis. To support the simulation of the fire, a team of HRT went to the preselected location carrying smoke bombs, torches (that produce a lot of smoke) and a smoke generator, to artificially generate a large smoke column that could be detected by the installed cameras from quite a distance. The AI software detected the smoke cloud and automatically triggered a fire alert in the dashboard. The team in the C3 verified the alert through the direct visualisation of the images taken by the camera.
View from camera A in the system
Fire simulation field trials
Following the fire incident detected earlier, bilateral communications between first responders on the field and chief operator in the C3 room are of high importance. The SAFERS IS in this case is involved through the Dashboard and Chatbot messaging functions. The dashboard is used by the C3 operators and the chatbot by the first responders on site, as expected, to allow continuous communication flow.
First responders using SAFERS Chatbot
Following the start of the fire, the head of operations in the C3 aimed at performing an initial fire propagation forecast to determine the potential impacts.
To that end, SAFERS on-demand fire propagation simulation was used to predict the potential fire spread and behaviour and assess their possible impacts. The simulation was conducted using real local weather and fine fuel moisture content data, from summer conditions and using the preselected fire ignition point. Different types of fire behaviour and spread simulation maps were generated and visualized for risk assessment purposes: hourly-based fire propagation throughout the following 8 hours, min and max Fireline intensity and min and max rate of spread. These maps offer a great tool for supporting decision making by identifying critical locations and potential impact.
Chortiatis fire simulation propagation
In the next step, a mission was created in the SAFERS IS’s dashboard and was assigned to the first responders on the field. Responders, using the chatbot, reported back their availability and accepted the mission. The same procedure was used by people assuming the role of citizens to demonstrate how citizens can provide information to the decision makers. The data shared from the C3 to the field was about the weather forecast, possible fire spread, images, etc. From the field to C3, the information shared was images, videos, messages, and fire behaviour from the field.
During a fire incident, continuous risk monitoring is essential. To that end, the monitoring of weather conditions and changes that may have an impact on fire behaviour should be carried out in the scenario. To that end, SAFERS IS was used to provide short-term weather forecast maps. Accordingly, personnel in the C3 consulted the layers corresponding to the main variables of the required weather data (temperature at 2m height, relative humidity, and wind direction and speed) to acquire the required information.
The SAFERS IS's crowdsourcing service was then used to search for and bulk filter fire-related Twitter messages. The parameters used to search and filter information from Twitter/X using the respective tool were the event's duration, the desired language (Greek and English), the kind of hazard (fire), and the area of interest. The system provided all relevant tweets from the particular location presented on a map. For the demonstration in Thessaloniki, and since there were no data for the pilot area (Seich Sou), data from another fire to assess the crowdsourcing service were used (i.e. tweets from the fire in Fokida Sernikakio that took place in July 2022 were used.
Sernikaki Fokidas wildfire 2022 tweets (crowdsourcing)
The on-demand fire simulator of the SAFERS IS was used to obtain information about the potential fire spread and behaviour with the specific input data provided. A second simulation was run, but specific fire combat actions were included in the fire propagation tool to assess their impact.
To illustrate the tool's capability, the C3 operators created the perimeter of the fire area as it would be at this stage and added specific fire suppression strategies throughout a defined time period. The system then generated an hourly-based propagation module for the given time span.
2.3. Exercise scenes III: Restoration and Adaptation
During the restoration and adaptation phase, the module of fire delineation and burnt area mapping was used to automatically obtain the shape of the region affected by the fire. This allowed an additional investigation of the generated damages and restoration procedures to be implemented in response.
To demonstrate the tools use, the C3 operator inserted specific data, such as the dates of the fire and the required resolution over a selected location on a map. The system then provided geospatial information with the requested output.
Rhodes island wildfire 2023 burnt area delineation
The final step in the recovery procedure is to monitor vegetation/plant restoration in the burnt regions. This allows decision makers to understand the fire's behaviour since they need to know how the soil can recover and how plantation & flora can influence fire evolution and behaviour. To demonstrate the post-fire monitoring in SAFERS dashboard, the post-fire vegetation recovery monitoring service was used to evaluate the evolution of the vegetation overtime during a period of 2,5 years after a fire event. The output data were maps to assess vegetation and soil recovery.
Loutraki Korinthias wildfire post monitoring
3. The feedback collected from the stakeholders
Following the demo, a feedback session was held with questions and the collection of first impressions, opinions, and recommendations from the audience. After that, the attendees were asked to reply to the feedback questionnaire about the use of the dashboard and the chatbot.
This feedback reaffirms the significance and value of the system's services in the context of fire management. In summary, the feedback for the dashboard can be deemed positive, and this is particularly noteworthy considering the presence of key stakeholders such as the Fire Service and Civil Protection Authorities from Greece. This positivity is evident in the suggested improvements put forth by participants, all aimed at enhancing the system's functionality and the information it provides.
SAFERS partners and participants during the Thessaloniki, Greek Pilot
As with all prior pilots, the demonstration's activities were aimed to test and provide feedback on the technologies and tools produced, directly involving operators and people to get direct input on the maturity, efficiency and ease of use.
Specifically, this demonstration attempted to achieve the following objectives: i/Showcase the current and potential use of the β2 version of the SAFERS Intelligent Services for operations related to wildfire risk management; ii/Test and validate SAFERS Intelligent Services in real-case scenarios through a demonstration scenario inspired by the Seich-Sou fire in 1997. However, since no data from that fire were available, the system would be validated against data from various fires that toom place in Greece to demonstrate even further its applicability and use; iii/Collect feedback for the improvement and fine-tuning of the system toward the final pilot in Greece; and iv/Enable networking and mutual acquaintance between SAFERS partners, Greek research and operational organisations, and other parties invited to the pilot.
The achievement of these objectives is part of an iterative process jointly conducted on the one hand by the SAFERS technical partners, who develop and progressively fine-tune capabilities and features for their ISs as part of the integrated SAFERS platform, and on the other by the pilot leaders, who coordinate the preparation and execution of operationally oriented scenarios to test the SAFERS ISs.
Learn more about the demo: Link