A New Drone Methodology for Accelerating Fire Inspection Tasks

Abstract

This study presents a validated drone-based methodology for inspecting fire protection belts in Galicia, Spain, with a focus on secondary protection belts surrounding settlements. Current manual inspection methods are limited by resource constraints and inefficiency, especially given Galicia’s steep slopes and fragmented, vegetated terrain. Our integrated approach combines high-resolution drone imagery, RTK positioning, GIS tools, and the Time2Parcel algorithm, enabling synchronized, parcel-level documentation at cadastral scale and allowing office-based technicians to directly review automatically generated video segments specific to each parcel for inspection verification. The methodology employs a hybrid classification system: automated assessments via orthophoto and LiDAR analysis and manual verification for cases with low confidence scores. Government technicians can perform office-based reviews without GIS expertise; the system automatically matches video to cadastral records, eliminating manual video review. Key results include the Time2Parcel algorithm for automatic video-to-parcel correlation, completion of inspections for 4934 parcels, and an operational efficiency increase of 68–70% reduction in inspection time compared with traditional methods. This workflow enables faster, safer, and more accurate inspections in highly fragmented rural contexts, improving legal compliance and environmental management.

1. Problem Description: Context and Legislation

During the summer, Galicia faces frequent forest fires, intensified by land fragmentation and rural abandonment, which have fostered the spread of pyrophytic species and shrubland. To mitigate fire risk and damage, the Xunta de Galicia implemented protection measures, notably establishing fire protection belts as per Law 3/2007, April 9th, on prevention and defense against forest fires in Galicia [1]. Article 20 of this law mandates landowners to maintain these belts clear, with the government empowered to intervene if necessary, and provides technical and operational criteria for their management.

Primary Protection Belts: Aligned with linear infrastructure (e.g., highways, railways, power lines), these belts act as barriers shielding critical utilities from fire.

Secondary Protection Belts: Defined as 50 m buffer zones surrounding populated and built-up areas, their purpose is to safeguard people and property from encroaching wildfires.

These secondary protection belts, which are the main focus of this work, create protective zones around urban areas, rural communities, development sites, and buildings intended for human occupation. This includes isolated houses, residential developments, landfills, campsites, gas stations, and industrial parks and facilities located within 400 m of forested areas.

Among other benefits, these secondary networks seek to prevent and minimize human and material damage by preventing fire from reaching buildings and population centers. However, for the successful implementation of these measures, the administration took on the task of manually inspecting the lands affected by these protection belts and issuing the necessary notifications to property owners. Furthermore, there was an intention to establish an automatic method for identifying the various tasks to be carried out on each plot so that once the inspection process was completed, all the tasks could be grouped for each parish or municipality.

Although primary protection belts, namely those lands adjacent to highways, expressways, railway infrastructures, power transmission and distribution lines, etc., are areas with more linear development and are more accessible, making inspection and control simpler, these secondary belts present various challenges.

• Population dispersion and multitude of areas: Galicia has countless rural settlements, making it by far the region in Spain with the highest number of population centers. In addition to this, one must consider the multitude of small clusters of buildings and isolated structures that have developed outside these established settlements.
• Irregular-shaped areas: The outline of these settlements and isolated buildings would theoretically form protection belts resembling concentric rings or rings with central holes. However, due to the unique and heterogeneous nature of Galician settlements, these protection zones actually form highly irregular and non-uniform shapes.
• Predominantly small landholding areas: This refers to the surroundings of areas where there is generally a high fragmentation of land, with small-sized properties and a very irregular layout of the parcels.
• Difficulty in delimiting parcels: The existing abandonment in many of the lands, as well as the large number of small plots (small-scale land ownership), and the absence of landmarks and boundary elements that define them greatly complicates the inspection process and identifying which property is in violation.
  Furthermore, reconsidering cadastral boundaries on the ground is deemed unfeasible considering the number of plots, the technical condition of the land (presence of dense vegetation hindering access, inability to use high-resolution GPS due to tree cover), and the economic factors (number of plots, need for preliminary work, slow process).
• Difficulty in delimiting on ground belts: Similar to the previous case, establishing the boundaries of the areas to be conditioned (belts) on the ground is also complex, facing the same issues as outlined in the previous point. Additionally, considering that this delimitation does not follow natural or traditional boundaries adds to the challenge.
• Difficulty of access: Many of the properties are difficult or impossible to access.

All these circumstances make on-site inspection, carried out manually, complex and often subject to estimations or approximations. Additionally, it may be prone to certain inaccuracies and entail high economic costs. Furthermore, these inspections must be completed before the fire season begins, creating significant time constraints that further complicate the process.

Due to these difficulties, the use of drones is proposed as a tool to inspect and detect breaches in secondary protection belts. This approach integrates drones with Geographic Information Systems (GIS) that incorporate custom-developed software specifically designed to optimize these inspection tasks. This includes a novel algorithm for correlating video footage with parcel boundaries and automated classification systems.

This use of drones for inspection purposes is not novel and has already been employed in other sectors, as discussed further in the following section.

2. Background and Current Approaches in Forest Fire Inspection

The application of drones, also known as Unmanned Aerial Vehicles (UAVs), in forestry has expanded significantly in recent decades, demonstrating substantial economic and operational benefits over traditional platforms in terms of repeat rate, spatial resolution, and accuracy. With their ability to provide real-time data, drones have become an indispensable tool in the protection and sustainable management of forests, offering rapid response capabilities for various forest management challenges [2].

The integration of drone technology with advanced sensors, Geographic Information Systems (GIS), and artificial intelligence has revolutionized forestry applications, especially in areas where traditional methods face limitations due to terrain complexity, accessibility issues, or the need for rapid assessment. This section explores the primary applications of drone technology in forestry with a focus on fire prevention and management [3].

2.1. Drones in Forest Fire Detection and Early Warning Systems

Early detection of forest fires represents one of the most critical applications of drone technology in forestry management. Drones equipped with specialized thermal cameras and sensors can detect fire outbreaks at their earliest stages, often before they would be visible to traditional ground-based or satellite monitoring systems [2,4].

Recent advancements include the development of systems capable of detecting thermal anomalies with remarkable precision, where drones can identify fire signatures from vibrations lasting as little as 500 milliseconds. These systems typically employ a combination of multispectral imagery, thermal sensors, and artificial intelligence algorithms to distinguish between normal environmental heat patterns and potential fire outbreaks [4,5].

The effectiveness of drone-based early warning systems stems from their ability to patrol large forest areas systematically and with greater frequency than traditional methods [6]. When integrated into broader monitoring networks, these systems can transmit real-time alerts to emergency response teams, significantly reducing reaction time and potentially preventing catastrophic fire spread. For instance, some advanced systems can detect fires within the first 12 h of ignition, which is considered the most critical window for effective containment.

2.2. Drones in Forest Fire Prevention and Risk Management

Beyond detection, drones play a crucial role in fire prevention through comprehensive risk assessment and fuel management monitoring [7]. By collecting detailed data on vegetation density, moisture content, and the presence of pyrophytic species, drones enable forest managers to identify high-risk areas that require preventive intervention.

Drones equipped with LiDAR (Light Detection and Ranging) technology can generate precise three-dimensional models of forest canopies and understory vegetation, allowing for accurate assessment of fuel loads and potential fire pathways. This capability is particularly valuable for planning controlled burns and other fuel reduction strategies that are essential for preventing catastrophic wildfires [7].

Furthermore, drone-based monitoring can track the effectiveness of existing fire prevention measures over time, providing data on vegetation regrowth patterns, the establishment of firebreaks, and the overall health of forest ecosystems. This continuous monitoring enables adaptive management approaches that can respond to changing environmental conditions and emerging threats [7,8].

2.3. Drones in Wildland–Urban Interface Management

The wildland–urban interface (WUI) presents unique challenges for fire prevention and management, as it combines natural vegetation with human settlements and infrastructure. In these critical zones, drones offer significant advantages for inspection and monitoring of protection belts and buffer zones mandated by regulations such as those established in Galicia under Law 3/2007 [7].

Drone technology enables precise mapping of these interface areas, identifying properties at risk and assessing compliance with vegetation clearance requirements. The high-resolution imagery captured by drones can reveal encroachment of vegetation into designated buffer zones, presence of combustible materials near structures, and other potential hazards that might not be readily apparent from ground-level inspections [9].

Studies have shown that fires in the WUI exhibit distinctive characteristics, including abundant and continuous combustible materials, complex environments, numerous ignition points, and rapid spread patterns that make firefighting particularly challenging. By monitoring these areas regularly with drones, authorities can ensure preventive measures are maintained and identify potential hazards before they contribute to catastrophic events [10].

Moreover, drones can assist in post-fire assessment of WUI areas, documenting damage patterns and evaluating the effectiveness of preventive measures, which provides valuable data for improving future protection strategies. This continuous cycle of monitoring, assessment, and adaptation is essential for building community resilience to wildfire threats in these vulnerable interface zones [10].

While the previous sections have demonstrated the significant potential and benefits of drone technology in forest fire prevention and management, the practical implementation of drone-based systems is not without obstacles. Understanding these challenges is crucial for developing effective and sustainable drone-based inspection methodologies.

2.4. Current Manual Inspection Practices and Limitations

While drone technology offers significant potential for forest fire inspection, as demonstrated in the previous sections, current inspection practices in regions like Galicia still rely heavily on traditional manual methods. Understanding these current approaches and their limitations is essential to appreciating the need for more advanced technological solutions.

Currently, the inspection of fire protection belts is carried out through in-person field visits conducted by human technicians who must physically access each required area within the designated zones, covering all population centers that need to be inspected. In Galicia, this process is particularly complex due to the region’s rugged geography and the evident abandonment of rural areas. This geographical complexity results in a significant number of inaccessible paths that have become overgrown with vegetation or deteriorated due to weather conditions, as well as the inability to access certain areas that require inspection due to excessive vegetation growth.

The temporal constraints of this manual approach present additional challenges. The entire inspection process—including communication with property owners and cleaning of the plots (whether performed by the owners themselves or by the Xunta de Galicia)—must be completed between February and May of each year. This timeframe is critical because the plots must be cleared prior to the wildfire season, as this represents one of the primary measures for protecting the population. Consequently, inspection periods must be both short and highly efficient.

Given these constraints, current inspections are carried out by numerous engineering teams and field workers using available resources. However, inspections are conducted only through sampling approaches, meaning that not all areas are inspected; instead, a percentage is selected based on various criteria and statistical methods. Efforts have been made to improve this process by incorporating satellite imagery (orthophotos), but this approach faces significant limitations. While these images can reveal vegetation density, they make it very difficult to estimate vegetation height, assess the condition of plots beneath tree canopies, and identify specific pyrophytic species.

Some field teams have experimented with drone technology to assist in inaccessible areas, demonstrating that drones can support inspection teams. However, these efforts have been limited by operational constraints, including the need for trained pilots, suitable environmental conditions, and equipment for intensive use (such as extra batteries and spare parts), making widespread deployment unfeasible throughout the entire inspection campaign. Additionally, this limited drone usage generated large numbers of videos that lacked specific references to individual plots under inspection. As a result, these videos served only as supplementary support for field teams during specific moments in certain forested areas, rather than providing systematic plot-by-plot analysis.

Current manual inspection methods suffer from significant limitations, including resource intensity, time constraints, limited coverage, accessibility challenges, and inadequate data organization. These limitations demonstrate the urgent need for more efficient and scalable approaches. The integration of systematic drone technology with Geographic Information Systems, as presented in the following section, offers a promising solution to address these fundamental challenges while maintaining the accuracy and compliance requirements essential for effective fire protection belt management.

2.5. Challenges in the Use of Drones

The use of drones is accompanied by several challenges that need to be addressed for their seamless integration into various industries [11]. While drones offer significant potential for revolutionizing various fields of inspection, they also face several challenges.

2.5.1. Technical Challenges

The implementation of drone-based inspection systems for secondary protection belts faces several operational and technological limitations. These technical challenges can impact the efficiency and effectiveness of drone deployment for large-scale environmental monitoring. The following section examines key constraints, including limited flight time and range, weather dependencies, and sensor limitations, that must be addressed when developing drone-based inspection protocols.

• Limited flight time and range: Current drone batteries limit flight time and range, restricting the area that can be inspected in a single mission [12,13]. Additionally, the type of drone and its energy efficiency play a significant role in determining its range, with specialized long-range drones being developed for professional purposes such as aerial photography and surveying. The flight time of drones is primarily restricted by the capacity of their batteries, with consumer-grade drones offering an average flight time of approximately 20 to 30 min. While recent technological advancements have allowed for longer flight distances, challenges persist in terms of battery life and range.
• Weather dependence: Adverse weather conditions such as high winds, precipitation, and extreme temperatures can adversely affect drone endurance, control, and safety, limiting their flyability and compromising data quality [12,14].
• Sensor limitations: Drones are equipped with various sensors that enable them to gather data for different applications. However, these sensors come with certain limitations that can impact their effectiveness in specific scenarios. Sensor resolution and accuracy may not be sufficient for detailed inspections in all cases [15]. Additionally, drones have a restricted payload capacity, and adding heavier or larger sensors may affect flight time and stability. Furthermore, adding bigger or heavier sensors can impact the drone’s aerodynamics and stability.

2.5.2. Regulatory and Legal Challenges

Implementing drones for protection belt inspections requires navigating a complex landscape of regulatory and legal considerations. These challenges create significant hurdles that impact operational capabilities and deployment strategies. The following section examines these key challenges.

• Air traffic control regulations: Integrating drones into airspace shared with manned aircraft presents significant safety and regulatory obstacles. Among these challenges are airspace congestion, collision risks, operational complexities, communication and control issues, surveillance concerns, and the necessity for robust regulations and technologies. The presence of drones alongside manned aircraft increases the likelihood of conflicts, particularly in densely populated airspaces like those near airports [12].
• Privacy concerns: The data collected by drones can raise privacy concerns, necessitating clear regulations and ethical considerations. Historically, drone regulation has primarily focused on safety, but there is an increasing need to address privacy issues. The enhanced capabilities of drone-based technologies raise questions about privacy and data protection laws. Drones have the potential to capture personal data, and their use must adhere to principles such as legality, fairness, transparency, purpose limitation, and information security to protect individuals’ privacy [13].
• Data security: Data security in drones is a critical consideration, given their role as data capture devices. Drones are vulnerable to cybersecurity threats such as hacking, unauthorized access, and data interception. If malicious actors gain access to a drone’s systems, they may be able to intercept data in real-time or access stored data, posing risks to privacy and security [16].

2.5.3. Operational Challenges

The following section addresses the significant operational hurdles organizations must overcome when adopting drone-based inspection systems, including equipment and training costs as well as the complexities of integrating drone data into existing operational frameworks.

• Cost of equipment and training: Acquiring drones and sensor equipment involves significant upfront costs. High-quality drones with advanced features can be expensive, especially those designed for specific applications such as aerial photography, surveying, or industrial inspections. Additionally, specialized sensors and payloads tailored to specific tasks can further add to the initial investment. Furthermore, properly trained personnel are essential for safe and effective drone operations. Investing in training courses, certifications, and ongoing education for personnel adds to the overall cost of drone operations. Equally important are the costs associated with obtaining necessary permits, licenses, and insurance coverage, including administrative tasks and ongoing regulatory compliance efforts.
• Integration with existing workflows: Integrating drone data into existing programs isn’t a straightforward plug-and-play endeavor. Achieving seamless utilization of this technology demands both dedication and flexibility. It entails re-evaluating existing workflows, fortifying data management systems to accommodate the influx of aerial imagery, and providing specialized training for personnel in drone-specific data analysis and interpretation. The endeavor extends beyond technical challenges; it necessitates a cultural shift within the organization to fully embrace this new technology, fostering collaboration among drone pilots, data analysts, and maintenance personnel. Additionally, organizations must implement robust data management systems to efficiently store, process, and analyze the data, addressing concerns related to storage capacity, data security, and data quality assurance [15].

Despite these challenges, the limitations of current manual inspection methods—particularly in complex terrains like Galicia—make drone technology a compelling solution when properly implemented. The time efficiency gains (reducing inspection time from weeks to days), improved safety for personnel, enhanced data accuracy, and the ability to access previously unreachable areas justify addressing these technological and operational challenges. Furthermore, many of these challenges can be mitigated through careful methodology design, appropriate equipment selection, and integration with existing workflows. The rapid advancement in drone technology, decreasing equipment costs, and evolving regulatory frameworks continue to improve the feasibility of drone-based inspection systems.

To better understand how these challenges have been addressed globally and to contextualize the Galician approach, it is valuable to examine international practices in automated fire protection belt monitoring.

2.6. International Comparison: Automated Methods and Buffer Zones

The implementation of buffer zones for wildfire prevention is a common strategy in many fire-prone countries. While legislation in Galicia uniquely defines “primary” and “secondary” fire protection belts—secondary belts being 50 m buffers around settlements—the broader concept is widespread internationally, as defensible space, fuel breaks, or wildland–urban interface (WUI) zones [9,17,18].

2.6.1. International Automation: Drones and Satellite Data

Recent years have seen a rapid global shift from manual supervision to semi-or fully-automated approaches combining drones, satellites, remote sensing, and GIS:

• Drones for inspection and early detection: Many countries have implemented UAV-based systems for both real-time fire detection and compliance monitoring. For example, UAVs equipped with thermal and multispectral sensors provide rapid and accurate hazard mapping over large, inaccessible territories [2,3,4,6]. Drones are also used for systematic surveillance of buffers/WUI zones, particularly in highly fragmented or peri-urban areas [7]. However, these efforts mostly focus on large contiguous areas, and rarely reach parcel-by-parcel documentation at cadastral/legal level.
• Satellites for regional monitoring: Satellite platforms (e.g., MODIS, Sentinel) are widely used for regional fuel mapping, fire detection, and risk prediction [19,20]. Such systems are effective for detecting fire outbreaks and monitoring overall hazardous conditions but generally lack the spatial detail required for micro-scale compliance or enforcement.
• WUI buffers and legal definitions: The concept of WUI and buffer enforcement is broadly established in North America, Australia, and parts of Europe. While the U.S. and Australia impose minimum clearance widths (30–100 m) and periodic vegetation management, the fine-grained cadastral approach used in Galicia to identify and document each parcel is uncommon [17,18].

While these international approaches demonstrate the widespread adoption of automated systems, the Galician methodology addresses several gaps in current practices, particularly for legally binding, cadastral-scale supervision.

2.6.2. Differential Aspects of the Galician Approach

The present Galician approach stands out due to its administrative and legal granularity, automation capabilities for fragmented landscapes, and a hybrid verification process. By leveraging high-resolution drone imagery, detailed parcel overlays, and the Time2Parcel algorithm, inspection data can be directly linked to cadastral records, which uniquely facilitates legal enforcement and reporting. Unlike many international systems that operate at broader landscape or wildland–urban interface scales, this methodology is especially distinctive in its ability to deliver per-parcel documentation, making it well suited for highly fragmented rural environments. Moreover, the inspection process combines automated classification with human expert validation, effectively addressing the challenges of fine-scale land use management, an approach that is often recognized but seldom implemented operationally elsewhere [2,7].

While automated UAV and remote sensing solutions are widely adopted internationally for wildfire prevention and inspection, their integration for legally binding, cadastral-scale supervision—as demonstrated in Galicia—remains rare and may serve as a best practice example for similarly structured regions.

Given these identified limitations in current international approaches and the unique challenges of Galician terrain, the following section presents our comprehensive drone-based methodology that addresses these gaps while leveraging the demonstrated benefits of UAV technology for fire protection belt inspection.

3. New Methodology for the Use of Drones in Forest Inspection

This section presents a detailed description of the new methodology and software developed for drone-based inspection tasks in forest areas. The main objective is to detect violations within secondary belts while considering the unique environmental characteristics and the specific legislative framework in Galicia. The methodology is tailored to ensure compliance with regional regulations and to adapt to the particularities of the territory, enabling a more efficient and accurate monitoring and enforcement process. In addition, we have developed specific algorithms that allow civil servants—without prior knowledge of video editing or geographic information systems—to perform inspection tasks from the office, thereby reducing costs and improving processing times.

The proposed methodology integrates geospatial data acquisition, processing, and field verification as interconnected steps to systematically assess biomass management and identify constraints affecting the studied plots. This structured approach enables comprehensive analysis and ensures reliable results. To enhance spatial accuracy and regulatory compliance, the methodology leverages advanced Geographic Information Systems (GIS) tools and high-resolution geospatial data. Additionally, we have incorporated software that allows inspection activities to be conducted independently of the physical location. Specifically, the system enables the tentative classification, validation, and documentation of each plot directly from the office. It also allows users to access the exact seconds in the video where a particular plot appears—and, conversely, to determine which plots are visualized at any given second of the footage.

The diagram in Figure 1 illustrates the proposed methodology. Before detailing each step, we first present an overview of the custom software that enables this workflow, followed by the detailed methodology in subsequent sections.

3.1. Software Overview and Role in the Workflow

The methodology is supported by custom-developed software that streamlines the inspection process and enables office-based review without requiring specialized GIS expertise from end users. The software serves as the central coordination tool that links drone-captured video footage with specific cadastral parcels, supporting both automated analysis and manual verification workflows.

The main functions of the software include (1) correlating drone flight paths and video timestamps with parcel boundaries to identify when specific properties appear in the footage, (2) providing a user-friendly interface for technicians to review flagged parcels and assign compliance classifications, (3) generating structured inspection reports that integrate spatial data with regulatory requirements, and (4) managing the workflow from raw drone data through final documentation.

What makes this software novel is its ability to automatically synchronize video content with cadastral records, eliminating the need for inspectors to manually search through hours of footage to find relevant parcels. Instead of using generic GIS tools that require extensive training, the software presents a streamlined interface where users can click on a parcel and immediately access the corresponding video segments. This direct video-to-parcel linking significantly reduces the time and expertise needed for thorough inspections.

The software fits into the overall workflow by processing the outputs from drone flights and orthophoto generation, then feeding organized data into the field inspection phase. It handles the complex spatial calculations behind the scenes while presenting inspectors with a simple point-and-click interface for parcel review and classification. A schematic representation of this integration within the broader methodology is shown in Figure 1.

This approach streamlines office-based review compared with generic GIS tools by providing purpose-built functionality for fire protection belt inspection, reducing the learning curve for administrative staff, and improving consistency across different inspection teams.

With this software framework established, we now detail the step-by-step methodology illustrated in Figure 1. The following sections describe each phase of the process, from initial data preparation through final reporting, showing how the software integrates with field operations to deliver comprehensive inspection results.

3.2. Delimitation of Belts and Parcel Identification

This step focuses on defining the boundaries of protection belts, which are critical for accurately analyzing the area of interest. The process leverages geospatial data provided by the local public administration in SHP format (Shapefile-a standard GIS vector data format) and cadastral systems to ensure precise and reliable delineation. The protection belts studied were georeferenced to guarantee accurate spatial localization.

Our software, integrated into the QGIS 3.28 tool, processes all these protection belts for a specific municipality according to the inspection parameters. The tool merges information from the belts with processed layers containing cadastral data (which in Spain relates a property to its owners, including its graphical definition). It also incorporates segmentation according to inspection processes (for example, the drone flight route map and flight planning), as well as metadata from other data sources and previous inspections when available.

The process consists of the following stages:

• Overlaying Protection Belts onto Cadastral Maps: The identified protection belts are superimposed onto cadastral maps to segment them into manageable units. This integration provides a comprehensive layer of spatial information, including property boundaries, ownership details, and land use data, which facilitates subsequent steps such as orthophoto generation and field inspection.
• Segmentation into Parcels and Plots: Through this overlay process, each belt is divided into specific parcels or plots. All relevant cadastral and spatial information is included, and new fields are added to the dataset to record inspection results and constraints identified during the analysis. This segmentation produces a structured parcel map that serves as the foundation for further operations.
• Data sources: The incorporation of terrain data was utilized to obtain the NDVI index [19,20]. This data was processed according to parameters specific to Galicia and its vegetation typology. The processed information was then integrated as metadata into a dedicated layer, which will subsequently enable more automated decision-making regarding the types of work to be carried out.
• Output Generation: The final output is a detailed, georeferenced layer containing all parcels, belts, and relevant metadata. This layer is essential for integrating additional geospatial and inspection data in subsequent steps.

3.3. Orthophoto Generation

This stage of the methodology focuses on acquiring high-resolution imagery. The process consists of the following steps:

• Imagery acquisition: Our team utilized both fixed-wing and rotor-based drones, each selected according to the specific needs of the terrain and inspection objectives. Two RPAS platforms were employed: an eBee equipped with a senseFly S.O.D.A 20 MP camera, and a DJI Phantom equipped with a 12 MP camera and 4K video recording capability. For the DJI Phantom 3 Professional with the best-performing battery available for this model was used, namely the DJI Intelligent Flight Battery (LiPo 4S, 15.2 V, 4480 mAh, 68 Wh, 365 g), which have ensured maximum flight autonomy (for us was around to 25 min per battery under optimal conditions) and high operational reliability.
  All platforms were equipped with high-definition cameras and Real-Time Kinematic (RTK) positioning systems, enabling the capture of high-resolution aerial imagery with sub-centimeter spatial accuracy. Fixed-wing drones were primarily employed for covering large areas efficiently due to their extended flight autonomy and range, while rotor drones were used for more detailed inspections in confined or complex environments requiring greater maneuverability. These aerial platforms were operated by trained personnel following predefined flight plans designed to ensure systematic and comprehensive coverage of the target zones. The execution of each flight mission was carefully planned and adapted to meet the operational and regulatory requirements specific to this study.
  In this study, two types of UAV platforms were employed with complementary roles. The fixed-wing drone (eBee) was used to capture high-precision orthophotos at low altitude over large areas, maximizing coverage and ensuring the spatial accuracy required for cadastral-scale analysis. In the secondary belts areas, a multi-rotor drone was deployed to record detailed video segments of individual parcels. These videos, processed through the Time2Parcel algorithm, allowed office-based technicians to verify inspection results without requiring on-site visits. This dual-platform strategy combined the efficiency and productivity of fixed-wing flights for systematic mapping with the higher detail and flexibility of rotor-based systems for parcel-level inspection, aligning equipment selection with the operational needs of wildfire prevention in fragmented rural landscapes.
  In the context of inspecting large, fragmented, and vegetated rural areas such as those present in Galicia, the selection of a fixed-wing drone over a more economical and common multi-rotor alternative was primarily driven by the operational requirements of the survey. Fixed-wing platforms, such as the eBee used in this study, offer significantly longer flight autonomy and greater range, enabling the coverage of extensive territories in a single mission with fewer take-offs and landings. This results in enhanced operational efficiency, as substantial areas can be mapped with the necessary overlap for high-precision orthophoto generation without frequent battery exchanges or interruptions. By contrast, multi-rotor drones, while more maneuverable and cost-effective, are limited by shorter flight times and reduced payload capacity, making them less suitable for systematic, large-scale mapping tasks but advantageous for detailed inspections in small or confined environments. Thus, the research scenario directly influenced the choice of platform: fixed-wing drones maximized area productivity and data consistency, while rotor-based models were reserved for specific tasks requiring greater agility. Ultimately, diverse drone models present distinct application effects depending on the spatial scale, terrain complexity, and inspection objectives, underscoring the need to align equipment selection with the operational demands of each case.
• Image processing: The collected images were processed using photogrammetry software by our team to generate orthophotos. The collected images were processed using Agisoft Metashape (version 2.0) with a custom Python (version 3.12) script developed by our team to automate the photogrammetric workflow and generate georeferenced orthophotos. During this step, parcels were classified into seven distinct types based on the tasks required, namely: felling and clearing to be determined, mechanical felling and clearing, mechanical felling, clearing to be determined, manual clearing, mechanical clearing, and parcels with no breaches.
• Overlaying orthophotos onto parcel maps: The orthophotos were superimposed onto parcel maps created during the previous phase. This integration was performed to enhance the spatial representation of the area, allowing for a highly detailed assessment of features such as biomass density, plot boundaries, and vegetation characteristics, while also incorporating the classification of parcels into the seven categories defined (felling and clearing to be determined, mechanical felling and clearing, mechanical felling, clearing to be determined, manual clearing, mechanical clearing and parcels with no breaches). This ensures that the analysis aligns with operational needs and supports decision-making processes.
  This classification operates through a two-stage hybrid process: (1) automated preliminary classification using our software based on vegetation metrics derived from orthophotos and LiDAR data, and (2) manual verification and refinement by trained technicians using the Time2Parcel algorithm for parcels where the automatic classification confidence score falls below 85 percent. The 85 percent threshold was selected as an optimal balance between automation and quality assurance, ensuring that only cases with sufficiently high classifier certainty are accepted without review, while parcels with greater ambiguity or potential for error receive expert attention. These orthophotos provide a comprehensive visualization of the inspected areas, facilitating the identification of features and constraints such as variations in vegetation cover, accessibility issues, and irregularities in parcel boundaries.
  The Time2Parcel algorithm relates each parcel to specific short aerial videos, facilitating a more detailed inspection when necessary. The detailed procedure for this algorithm is shown in Algorithm 1. These short aerial videos provide a comprehensive visualization of the inspected areas, facilitating the identification of features and constraints such as variations in vegetation cover, accessibility issues, and irregularities in parcel boundaries.

Algorithm 1 Time2Parcel: Correlating Drone Video with Parcel Visibility. This algorithm filters usable video moments for each parcel by identifying when specific parcels are visible in the drone footage.

Require: target_parcel, drone_video, rtk_gps_log, camera_parameters Ensure: parcel_video_segments, quality_metrics   1:  Initialize visibility_intervals $\leftarrow \varnothing$   2:  for each video_frame in drone_video do   3:        frame_timestamp ← GetTimestamp(video_frame)   4:        drone_position ← InterpolateGpsPosition(rtk_gps_log, frame_timestamp)   5:        camera_footprint ← CalculateGroundFootprint(               position: drone_position,               altitude: drone_position.height,               fov: camera_parameters.field_of_view,               gimbal_angle: camera_parameters.gimbal_angle       )   6:        if SpatialIntersection(camera_footprint, target_parcel.geometry) then   7:             Add frame_timestamp to visibility_intervals   8:        end if   9:  end for 10:  video_segments ← ConsolidateTemporalIntervals(visibility_intervals) 11:  for each segment in video_segments do 12:        quality_score ← EvaluateSegmentQuality(segment) 13:        if quality_score ≥ MinimumQualityThreshold then 14:              Add segment to parcel_video_segments 15:        end if 16:  end for 17:  return  parcel_video_segments

The hybrid approach is necessary due to the complexity of Galician terrain and vegetation patterns and legal requirements. Fully automatic classification faces limitations in distinguishing between different pyrophytic species, assessing vegetation health under varying lighting conditions, and evaluating compliance in areas with complex topography or mixed land use. Manual verification ensures accuracy while maintaining efficiency for large-scale inspections and is required by the administration to allow direct access to the inspection results. In practice, it enables staff to refine the areas automatically classified by vegetation algorithms during office work, and subsequently facilitates the official review by the administration. These video segments provide direct evidence of the parcel condition, making it possible both to validate the automatic classification and to support the administrative decision process.

In our algorithm, the function EVALUATESEGMENTQUALITY(segment) is responsible for filtering the raw visibility intervals and selecting only those that meet a minimum standard of usefulness. This step is performed to accelerate the work of administrative staff, who otherwise would need to manually review long video sequences linked to a secondary belt, but in which the parcel is not actually visible. The function assigns a normalized score between 0 and 1 to each video segment, which reflects how well the target parcel is captured in the video. Segments that do not reach the minimum score are discarded, ensuring that the final dataset is both consistent and reliable.

The most influential factor in EVALUATESEGMENTQUALITY is the coverage of the parcel, estimated from the four reference rays projected through the centers of the image quadrants. For every frame (only one frame per second), we count how many of these rays fall inside the parcel polygon, giving discrete coverage values of 0, 0.25, 0.5, 0.75 or 1. Averaging across all frames of a segment provides a straightforward measure of whether the parcel is consistently visible. In our implementation, coverage is therefore not computed as an area ratio but as a discrete approximation based on these four rays, which act as representative samples of the camera footprint.

A second element integrated into the quality function is the viewing angle of the camera relative to the terrain and parcel. Segments recorded at moderate angles—close to nadir—are preferred, because they show the parcel with minimal distortion and more homogeneous resolution but in our flights, the camera is operated at an oblique angle of approximately 45°, which provides a wider field of view and allows vegetation height and structural characteristics to be perceived more clearly, offering additional information that is not visible from a strict nadir view. While this angle introduces more geometric distortion compared with near-nadir configurations, it was found to be a practical trade-off that balances visibility, coverage, and operational efficiency. As the obliquity increases, the score is progressively reduced, penalizing views where the parcel appears only from the side. This ensures that the algorithm does not only reward visibility but also favors geometries that make the parcel easier to interpret.

Two additional factors refine the evaluation: the margin between the parcel centroid and the boundary of the footprint, and the temporal stability of the capture. The first penalizes cases in which the parcel is projected very close to the image edge, since these are more prone to incomplete or cropped views. The second penalizes abrupt variations of parcel position across consecutive frames, which indicate unstable or blurred sequences. Together with coverage and angle, these factors provide a balanced description of the actual quality of each segment. In our implementation, the weights applied to each factor were a coverage of 0.60, an angle of 0.15, a margin of 0.15, a stability of 0.10. The resulting weighted score is compared against a single acceptance threshold set to 0.72. Segments with a quality score below this value are discarded, while those with a score equal to or above it are retained. The choice of 0.72 was made empirically after testing different flights: lower thresholds admitted segments with very partial or unstable views of the parcel, while higher thresholds excluded segments that still offered sufficiently clear visibility. The value of 0.72 was therefore selected as a balance point that reduces the workload of administrative staff without being excessively strict and losing useful material, while also acknowledging that inspectors prefer to have additional video segments available rather than relying solely on a smaller set of “perfect” recordings.

3.4. Digital Terrain Model (DTM) Construction

Due to the presence of dense vegetation on many plots, direct measurement of terrain elevation may not be feasible. To address this limitation, a Digital Terrain Model (DTM) is generated using LiDAR data provided by the National Geographic Institute (2nd coverage) [21].

A Digital Terrain Model (DTM) is a geospatial representation of the surface of the bare earth, created in this case by processing LiDAR data. The process involves acquiring a dense point cloud through LiDAR, where laser pulses capture elevation data by measuring the time taken for signals to return from the surface. The raw point cloud is then classified to isolate ground points from non-ground features such as vegetation and buildings. Using interpolation techniques like Triangulated Irregular Networks (TIN) or rasterization, these ground points are converted into a continuous elevation model. The DTM is further refined through error correction and smoothing to ensure accuracy. This model, represented as a raster or TIN, provides a high-resolution foundation for applications such as slope analysis, hydrological modeling, and terrain accessibility assessments.

3.5. Field Inspection

The field inspection phase is a critical step with two main objectives: determine the compliance of the parcel and the accessibility of each plot or parcel.

• Compliance of the parcel: A static image, although high resolution, often provides unclear diagnoses as it is difficult to determine the height of vegetation and the presence of pyrophytic species. To support the inspection process, a new algorithm, Time2Parcel, was developed to associate each parcel with specific short aerial videos captured from multiple points of view. Using drone-captured aerial video and the previously mentioned predefined flight path with RTK positioning information, this algorithm identifies each parcel in the video.

As shown in Figure 2, the Time2Parcel algorithm annotates the drone’s predefined flight path on the protection belt with relative time in video recording and overlays on the map with parcel boundaries.

The result of this process is a short collection of videos in which the parcel can be more easily evaluated, all of them extracted from the video recorded during the aerial image acquisition used to generate the orthophoto. This means that only one flight is required to gather the necessary data, thereby avoiding the need for additional field visits and the associated logistical challenges, time consumption, and operational cost.

• Accessibility of each plot or parcel: To determine whether machinery could effectively reach the area for maintenance or intervention tasks, the accessibility of each plot was evaluated. This was possible using the Digital Terrain Model to determine terrain slope as well as cadastral information of each parcel to identify entry points and boundary pathways.
  Field accessibility information is used to determine the kind of intervention recommended to clear vegetation and achieve compliance in the parcel. Usually, it is not possible to access the land with machinery if the slopes are too steep or there is no nearby entry path to the parcel, which means that inspection teams must be deployed on foot.

3.6. Assessment of Constraints

This stage integrates additional data layers to identify potential limitations or restrictions that may impact operational tasks, as shown in Figure 3. This step ensures compliance with legal, environmental, and operational frameworks while highlighting areas requiring special attention or interventions. The process consists of the following key components:

• Identification of legal constraints: Using cadastral data and relevant legislation, this step identifies areas with specific legal restrictions, such as protected zones, property disputes, or areas requiring administrative permissions for intervention. These constraints are cross-referenced with geospatial data to ensure accurate identification.
• Environmental considerations: The analysis incorporates environmental factors such as vegetation density, the presence of pyrophytic species, and proximity to natural resources. These elements are evaluated to prioritize areas that pose the highest risk of spreading fires or ecological disruption, paying particular attention to the unique characteristics of each type of belt.
• Archaeological and cultural restrictions: In regions such as Galicia, where historical and cultural sites are prevalent, this step includes verifying whether any parcels overlap with protected archaeological zones or areas of cultural significance, considering the specific context of each belt type.
• Accessibility challenges: Data collected from the Digital Terrain Model (DTM) and field inspections is used to assess terrain accessibility for each of the four belt types. Steep slopes, dense vegetation, or lack of entry points are documented as constraints that could hinder operational efforts, taking into account the distinct features of Prioritized, Rural core, Industrial park, and Police belts.

3.7. Data Integration

This final phase consolidates all collected data into a unified framework for comprehensive analysis and report generation. The workflow from raw data to final reports follows a systematic sequence of key processing steps:

• Raw data ingestion: Drone media files, positioning logs, and cadastral datasets enter the system with the purpose of establishing the foundational data inventory. This process generates organized data repositories that comply with standardized formats and structured metadata, providing the solid foundation necessary for all subsequent system operations.
• Spatial data processing: Orthophotos, terrain models, and parcel boundaries are generated and aligned to create georeferenced base layers intended for spatial analysis. This procedure produces precisely mapped cartographic layers along with summary statistics for each protection belt, establishing the essential geospatial framework for territorial assessment.
• Video–parcel correlation: Flight paths and timestamps are correlated with cadastral boundaries to identify parcel visibility windows, enabling targeted and efficient video review. The result is indexed video segments specifically linked to individual parcels, facilitating focused inspection of each property.
• Automated classification: Initial compliance assessments are generated using vegetation metrics and accessibility criteria, with the purpose of flagging parcels that require specialized human review. This automated process produces preliminary classifications accompanied by confidence scores that guide the prioritization of manual verification work.
• Manual verification: Trained technicians review flagged parcels using video segments and classification tools to ensure accuracy and regulatory compliance. This critical quality control stage generates validated parcel assessments and defines specific intervention requirements for each inspected property.
• Constraint integration: Legal, environmental, and operational limitations are applied to refine intervention plans, ensuring the practical feasibility of proposed actions. This integration process results in actionable recommendations accompanied by priority rankings that guide effective implementation of corrective measures.
• Report compilation: All parcel data, classifications, and recommendations are consolidated into structured inspection reports, providing the documentation necessary for administrative action. The output consists of standardized reports that meet regulatory requirements and facilitate decision-making by competent authorities.
• Quality assurance: Final validation ensures data completeness and consistency across all parcels and protection belts, maintaining established inspection standards. This stage produces quality-assured datasets ready for operational use and prepared to support critical decisions in territorial management.

In the proposed methodology, a systematic and multi-source data collection is performed using various sensors installed on the drones. Core data sources include high-resolution RGB cameras, multispectral sensors, and GNSS positioning systems. This combination allows for precise spatial mapping, vegetation cover analysis, species identification (including pyrophytic species), and estimation of variables such as vegetation height and terrain accessibility. Integrating these data types—enriched with georeferenced auxiliary information—enables objective, efficient assessment of the status of secondary protection belts, ensuring compliance with technical standards set by current legislation.

Having established the theoretical framework and technical methodology, we now demonstrate its practical application through a real-world case study in Galicia, showing how the system performs under actual field conditions and validates the proposed approach.

4. Practical Case

This section presents a practical example of applying the methodology to a real case in a region of Galicia. The following subsections will detail the implementation of the methodology and the results obtained.

4.1. Practical Case: Delimitation of Belts and Parcel Identification

Local regulations and environmental factors specific to Galicia were taken into account during this step, addressing challenges such as fragmented land ownership and irregular parcel shapes. These considerations are particularly important given the unique territorial and legislative characteristics of the region.

Figure 4 illustrates a map of the inspected areas. This study analyzed a representative region in Orense, Galicia, where the specific conditions—particularly the high risk of wildfires—underscore the importance of accurately defining and inspecting protection belts.

Figure 5 shows the process where the cadastral parcel layer of the area was overlaid with the delimitation layer defining each belt to be inspected. In the figure, green areas represent riverbeds, purple and pink areas show cadastral belts, and orange areas indicate cultural and protected patrimony zones. Subsequently, the initial vector layer was clipped, resulting in a new layer named PARCELAS.shp. This layer contains all cadastral information, along with additional fields created to store inspection data and other relevant information from the remaining layers.

4.2. Practical Case: Orthophoto Generation

This step is particularly significant in regions like Galicia, where fragmented land ownership and dense vegetation complicate traditional inspection methods. The high-resolution orthophotos offer a reliable visual aid for detecting areas requiring maintenance or intervention, while their integration with other geospatial data ensures a comprehensive and accurate assessment of the protection belts.

In this particular case study, a single RPAS system was utilized: the Ebee fixed-wing drone, equipped with high-resolution cameras.

• Fixed-wing system (Ebee): This equipment was employed to map the settlements and generate large-scale orthophotos. Ebee (AgEagle Aerial Systems Inc., Wichita KS, USA (previously senseFly. Switzerland)) is a long-range, high-endurance fixed-wing drone capable of covering extensive areas in a single flight. Depending on the camera configuration and flight conditions, it can map significant territories with sufficient lateral overlap to generate high-precision orthophotos. In this study, we employed the standard senseFly S.O.D.A 20 MP camera (AgEagle Aerial Systems Inc., Wichita, KS, USA (previously senseFly. Switzerland)).

Ebee is known for its long flight duration and ability to carry high-resolution camera payloads, making it suitable for large-scale mapping and surveying applications.

Figure 6 overlays the orthophoto onto the cadastral parcels, making the difference in biomass density and plot boundaries easy to plot.

4.3. Practical Case: Digital Terrain Model Construction

The construction of the Digital Terrain Model (DTM) was a critical step in addressing the challenges posed by dense vegetation and irregular terrain, which made direct elevation measurements impractical. This process leveraged LiDAR data provided by the National Geographic Institute (IGN), specifically the second coverage, which offered a point cloud density of 1 point/m² and a vertical accuracy of less than 20 cm RMSE Z.

The LiDAR point clouds were processed using geospatial software to generate a high-resolution DTM. The data adhered to the ETRS89 geodesic reference system with UTM projection and orthometric heights. The point clouds were automatically classified and colored using RGB data derived from orthophotos, with a pixel size ranging from 25 to 50 cm.

Using the DTM layer, a slope map was generated with the appropriate QGIS algorithm. For each parcel, the average, maximum, minimum, and most frequent (mode) slopes were calculated. This information was transferred to the parcels in the PARCELAS.shp layer.

4.4. Practical Case: Field Inspection

The inspection of each field was carried out using the output videos of the new Time2Parcel algorithm. Trained technicians review parcels flagged by the automatic classification system for manual verification, using the Time2Parcel algorithm to access specific video segments and make final determinations on compliance and intervention requirements. The technician also had to take into account the accessibility of each plot; he could not only determine vegetation status but also classify the type of intervention needed to be carried out. As shown in Figure 7, the result is a map with each parcel annotated with the kind of intervention needed, whether it is brush cutting or removing pyrophytic species trees. The figure employs a color-coded system to denote parcel inspection status: green indicates compliant parcels, orange designates parcels requiring supervisory review, and red represents non-compliant parcels. The pink, time-stamped markers pinpoint the exact video frames in which each parcel is visible, as identified by the Time2Parcel algorithm.

4.5. Practical Case: Assessment of Constraints

At this stage, the methodology integrates additional data layers to identify potential constraints. These constraints may include legal, environmental, or archaeological restrictions that could impact operations. The evaluation ensures compliance with regulatory frameworks and highlights areas requiring special consideration. The results of this step inform sustainable and responsible management strategies.

4.6. Practical Case: Data Integration

The final step consolidates all processed data into a unified dataset. This integration includes cadastral information, geospatial analyses, and results from field inspections and constraint evaluations. Historical data is also incorporated to provide context and support trend analysis. The resulting dataset enables comprehensive decision-making and facilitates the completion of templates required for operational planning.

5. Results and Discussion

The practical implementation of the drone-based methodology in Galicia, as detailed in the previous section, provides valuable insights that enable evaluation of the system’s performance and effectiveness in real-world conditions. This section presents and analyzes the key findings from our field deployment, examining both the operational outcomes achieved and the lessons learned through practical application. The analysis encompasses inspection outcomes and system capabilities, efficiency comparisons between drone-based and traditional manual approaches, scalability considerations for broader implementation, and practical factors affecting operational deployment. These results validate the proposed methodology while providing critical insights into the transformative potential of drone technology for fire protection belt inspection and similar environmental monitoring applications.

5.1. Inspection Outcomes

Using the methodology described above, numerous inspection tasks have been performed to identify and control violations. This approach, along with the implemented system, enabled the inspection and control of 4934 fields in just one year. Beyond ensuring spatial organization of the information, the system facilitated work with multiple layers, such as slopes of the terrain, altitude, vegetation (brushwood/trees), and field accessibility. This capability allowed for the determination of necessary operations and their execution methods, enabling the creation of nearly automated plans for each parcel to establish the required tasks effectively.

Table 1. Parcel data by region.

Metric	Cualedro	A Merca
Total inspected area (Ha)	71.5 Ha	50.5 Ha
Number of parcels	2780	1395
Average area (m2)	613	690
Area inside belt (m2)	256	362

presents the data obtained for each evaluated unit.

As shown in Figure 8, most parcels inside the Cualedro Secondary Protection Belt had an area below 100 m².

Table 2. Compliance of Plots in 2023 and 2024 (Cualedro).

Category	Total Number of Plots
Always non-compliant (2023 and 2024)	944
Compliant in 2023, non-compliant in 2024	222
Non-compliant in 2023, compliant in 2024	288

presents information regarding compliance and non-compliance across the different parishes within the municipality of Cualedro.

5.2. Efficiency Comparison: Drone-Based Vs. Manual Approaches

A critical aspect to address is the time required for conducting these inspections, as it directly influences the process’s efficiency and feasibility. By adopting the proposed drone-based methodology, substantial improvements can be made in workload optimization and reductions in personnel time demands.

Table 3. Time spent on tasks using the drone-based approach.

Task	Personnel	Time (A Merca/Cualedro)
Fieldwork	Pilot	3–4 days/2 days
GIS Technician work	IT Engineer	1 day/1 day
Inspection Tasks (2 people)	Agricultural engineer, Forestry engineer	3 days/3 days

provides a detailed summary of the specific time allocations for various tasks performed by the team members using our drone-based approach.

In contrast to the drone-based methodology, manual inspection processes would require significantly more resources. Based on data from previous years (the company had carried out manual inspections on the same plots during the previous three years, providing a consistent reference database for comparison), it can be estimated that for the same plots, approximately 200 plots per person per day could be covered in the best-case scenario. This would translate to around 25 days for one person or 13 days for two people working simultaneously. These figures assume the use of basic equipment, such as a handheld GPS device, which provides location tracking and minimal field data entry capabilities, as well as the ability to georeference photographs.

However, even with these tools, the manual approach has significant limitations. Once the field inspection is completed, the collected data would need to be organized and processed, adding additional time and effort to an already labor-intensive process. Furthermore, the manual approach is inherently constrained in terms of data collection. It would not be feasible to gather all the fields of information required for the scope of this project. For example, manually documenting 200 plots per person while also recording compliance status, observations, reasons for non-compliance, and taking georeferenced photographs of the plots to document non-compliance would be impractical, if not impossible, within the given time frame and resource constraints.

The results clearly demonstrate that the drone-based approach reduces inspection time from approximately 25 days for one person (or 13 days for two people) using manual methods to just 7–8 days total with our methodology. This represents a time reduction of approximately 68–70% compared with manual methods, while simultaneously providing more comprehensive data collection, higher accuracy, and better documentation of compliance issues.

5.3. Scalability Benefits

The proposed drone-based system offers significant advantages in efficiency and scalability. As the area to be inspected increases or the number of plots grows, the average time and costs associated with the drone-based approach decrease due to its ability to cover larger areas quickly and accurately. The drone system automates data collection and georeferencing, ensuring consistency and completeness while minimizing human error. Moreover, the real-time or near-real-time processing of collected data reduces the time needed for post-inspection analysis and organization.

This stands in stark contrast to the manual approach, where increased inspection areas or more plots result in proportional increases in time and costs. The inefficiencies of the manual process become increasingly apparent as the scope of the inspection grows, making it less viable for large-scale applications. Thus, the drone-based system not only provides a more efficient and cost-effective solution but also enables a higher level of data accuracy and comprehensiveness, making it a superior alternative for modern inspection requirements in agricultural and forestry management contexts.

5.4. Practical Implementation Considerations

While the technical results demonstrate clear advantages, practical implementation involves addressing operational logistics and data protection requirements. The workflow accommodates these challenges through flexible deployment schedules that work within weather constraints and regulatory windows, standardized data handling protocols that ensure privacy compliance when processing cadastral and video information, and modular training programs that enable administrative staff to operate the system without extensive technical backgrounds. These practical considerations are built into the methodology design, allowing public administrations to adopt the system while maintaining their existing administrative frameworks and meeting legal obligations for data security and citizen privacy.

The implementation of this methodology in Galicia reveals terrain-specific operational challenges that directly impact inspection efficiency. The region’s rugged geography, steep slopes, and dense vegetation create compound limitations beyond standard drone operations. Galicia’s territorial structure—characterized as ”by far the region in Spain with the highest number of population centers” with countless small, isolated buildings—required complex flight path optimization for highly irregular protection belt shapes. The small average parcel sizes (613–690 m²; with most below 100 m²) necessitated more intensive operational requirements, including frequent altitude adjustments and precise positioning that increased battery consumption compared with regions with larger, consolidated parcels. Additionally, the interaction between weather conditions and abandoned rural areas, where ”inaccessible paths have become overgrown with vegetation or deteriorated due to weather conditions”, compounds accessibility challenges during the critical February–May inspection window.

Based on these comprehensive results and the demonstrated operational benefits across multiple evaluation criteria, the following section synthesizes the key findings and presents the main conclusions of this research, along with recommendations for future development and broader implementation of drone-based inspection systems.

6. Conclusions

This article presented a comprehensive analysis of the use of drones as an innovative tool to assist public administrations in inspection tasks, particularly in the context of forest fire prevention in Galicia. The study focused on the deployment of drones for inspecting fire protection belts, evaluating their efficiency, accuracy, and cost-effectiveness compared with traditional manual methods. Our implementation successfully inspected 4934 fields in just one year, specifically 2780 parcels in Cualedro and 1395 in A Merca, demonstrating the system’s capacity to handle large-scale inspection tasks.

The integration of GIS allowed for the efficient processing and visualization of data collected by drones, including vegetation height, land accessibility, and the presence of fire-prone species. This approach enabled effective work with multiple layers, such as slopes of the terrain, altitude, vegetation (brushwood/trees), and field accessibility. By overlaying cadastral maps and belt boundaries, drones enabled the precise identification of areas requiring maintenance, which would have been challenging and time-consuming with traditional methods. This capability was particularly valuable given the small size of many parcels, with most in the Cualedro Secondary Protection Belt having an area below 100 m².

The findings underscore that drones can significantly enhance the efficiency of inspection processes, reducing the required time from approximately 25 days for one person (or 13 days for two people) using manual methods to just 7–8 days total with our drone-based approach. This includes 3–4 days of fieldwork by a pilot, 1 day of GIS technical work, and 3 days of inspection tasks by two engineers. Additionally, the high-resolution imagery and real-time data provided by drones enable more accurate assessments of fire protection belts, reducing human error and improving overall effectiveness. Our drone system also facilitated multi-year compliance monitoring, allowing us to identify a concerning 82.42% rate of persistent non-compliance across 944 plots in Cualedro between 2023 and 2024.

The main conclusion of this study is that drones can be effectively employed for the inspection of fire protection belts, offering a reliable and scalable solution to mitigate the risks associated with forest fires. Their ability to rapidly identify areas requiring intervention, coupled with their integration into GIS workflows, represents a significant advancement in forest fire prevention strategies. As the area to be inspected increases or the number of plots grows, the average time and costs associated with the drone-based approach decrease, in stark contrast to the manual approach, where increased inspection areas result in proportional increases in time and costs. By leveraging drone technology, public administrations can optimize their inspection efforts, reduce operational costs, and enhance the protection of natural resources, human life, and infrastructure.

Future research and development could focus on further automating the analysis of drone-captured data, improving battery life and payload capacities, and addressing regulatory and privacy concerns to ensure the broader adoption of this technology in diverse applications. Additionally, efforts could be directed toward improving compliance rates in persistently problematic areas such as those identified in our study. The potential for drones to revolutionize inspection processes extends beyond fire protection belts, offering valuable insights for environmental conservation, agricultural monitoring, and infrastructure management in regions characterized by fragmented land ownership and challenging terrain.