In this task advanced analytics algorithms ingest sensing and imaging data and provide a systematic solution for detecting and preventing fires. These algorithms will have the ability to explore the different kinds of data acquired (fire indices), while discovering and addressing new patterns (e.g. rule-based detection) leading to better decisions and faster reaction. The FDPR algorithms will be designed based on the needs, requirements and type of data reported in WP3 and will employ the data acquired (both sensing and image processing data) to provide real-time monitoring. More specifically, WP3 will provide data that will allow for the categorization (fire indicii), that in turn will allow for a risk categorization based on probability analysis. In particular the following parameters will be required, in order to provide a holistic approach: ▪ Land Cover – the biophysical characteristics (eg vegetation and type of it) ▪ Topography – aspect (the direction at which the slope faces), elevation ▪ Human Factors – proximity to residential zones, proximity to the road network Based on these, a five-point risk index (very low, low, moderate, high, very high) will be constructed. The FDPR mechanism will therefore determine the levels of risk of fire as well as determine possible locations of fire. In case of a fire detection, drones will be sent in situ to verify that there is an ongoing fire, thus providing an early warning system that will contribute to the active prevention and spreading of fires. Our algorithms will contribute towards the synchronisation of first response and emergency services by evaluating current reaction approaches and provide recommendations and supportive information steaming from our proposed FDPR mechanism. FDPR will be accessible to the user through the UI developed in WP3. This will be achieved through a unified API for the user interface as this will be defined in T3.3 and T3.4.

This task will develop a module for both offline illegal logging detection and real-time illegal logging and hunting detection. The former will mainly utilize satellite data for the calculation of spectral indices and spectral brightness characteristics to obtain information about illegal logging processes in disturbed areas. For analysing satellite images, several image processing techniques can be used such as image classification, spectral mixture analysis and Vegetation Indices, as well as the most commonly approach of using spectral vegetation indices (sVI) due to their ability to provide a spatiotemporal analysis (Task 5.4). For the latter, both audio data from acoustic sensors that will be deployed on the field as well as proximate images from drones will be used to identify illegal logging and hunting via AI techniques in order to trigger the 2-way verification protocol.

The module of this task will be responsible for identifying and recommending to the authorities / end-users possible areas for afforestation or reforestation. This requires the implementation of a multi-criteria analysis method (calculation of spectral indices and spectral brightness characteristics) on satellite images (Task 5.4) and/or proximate UAV images and machine learning (e.g., image classification, spectral Mixture Analysis and Vegetation Indices) for identifying and recommending sites. The intelligence module should be able, for example, at a first level, to identify the landscapes (spots) with low forest density (specifying the lowest density of forest request in Cyprus forests according to the forest type) and use a simulation process on trees density, slope, aspect and elevation of the site. At a second level, the modules should be able to evaluate elements of the ecological condition, such as: soil type, soil depth, soil texture classes and corresponding available water capacities. For example, the determination of forest canopy via satellite and/or UAV images will play an important role, since it acts as a regulator in forest ecosystems and it is a factor that affects the microclimate and the soil conditions. The density of the forest canopy is associated with the forest development and can be used as an indicator of forest degradation and consequently in the design and implementation of forest restoration programs. Finally, the module should be able to correlate all data from both levels in order to identify the locations for applying afforestation or reforestation programs. Unit testing will be performed upon the completion of the module in order to provide a bug-free module for the final integration in Task 7.1 of WP7.

This task will be responsible for developing all supportive mechanisms and various simple warning systems required for effective forest monitoring and management. Particularly, this task will develop the 2-way verification protocol that is needed in all intelligence modules of Tasks 6.1, 6.2 and 6.3. This protocol will be triggered in case an incident is detected (e.g., fire, illegal logging, etc.) in a specific geo-location. The latter will be automatically provided to the UAV that will fly to the area for verifying the incident either by providing real-time audio, video streaming and/or images. This will allow real human intervention only on verified cases. Additionally, simple rule-based algorithms will be developed for providing alerts/warning for various other incidents, pressures and impacts affecting sound forest management regime. For example, composite sensors and UAV proximate images will offer surveillance for overgrazing and other in-forest human activities such as camping and picnic sites, high ecological value sites, nature trails, illegal littering and dumping of wastes etc that will be identified in WP3. Moreover, composite sensors will provide data regarding the water, air, soil of the forest and in cases that either an undesired element is detected or an element passes specific thresholds provided by the forest experts then the platform will notify the corresponding authority. Finally, if the water flow in a forest falls below a particular limit, the authorities should be notified for changing the location of the animal watering cans in order to adequately serve the forest animals (e.g., Cyprus mouflon - Ovis orientalis ophion, which is protected based on EU 92/43 – Habitats Directive).