Drowning in a Sea of Data: The Hard Parts of Avian Hazard Management

Beca’s Jessica Tucker explores one of the hardest parts of Avian Hazard Management, what do organisations do with reams of data?

This article is the fourth and final part in our series exploring how cutting-edge technology can make everyday better for the Aerospace sector. Here's Part 1, 2 and 3.

In the world of aviation, where precision reigns supreme, the management of airborne hazards at low altitudes represents a complex and dynamic challenge. Birds, drones, and other unexpected visitors to airspace present risks that are both persistent and challenging to predict. Despite continuing advances in detection technology, these threats remain a costly and potentially catastrophic problem. This thought piece explores one of the hardest parts of avian hazard management: what does one do with all that data generated from aircraft hazard detection systems? We've drawn insights from research on detection systems and the evolving landscape of aviation safety.

The scope of the challenge
Birds, drones and to a lesser extent, bats, make up a trifecta of low-altitude hazards that test the limits of current detection and response systems. Bird strikes, the most studied of these threats, typically occur during critical flight phases like take-off and landing, with over 70% of them happening below 500 feet.

Migration patterns and seasonal behaviours amplify the unpredictability of these encounters. Meanwhile, drones — operated by untrained users and, knowingly or unknowingly in violation of airspace regulations — pose an increasing risk of mid-air collisions or disruptions to navigation systems.

Aside from the well-known, potentially catastrophic safety risks of bird strikes, the financial implications of these hazards are staggering. Bird strikes alone cause annual global costs exceeding AUD$3 billion, encompassing repairs, cancellations and operational disruptions. While less historically documented, recent drone incidents have been calculated to carry similar costs due to damage, cancellations and other factors. These figures underscore an urgent need for better tools and strategies to safeguard aviation operations.

Navigating data overload
One of the hardest parts of avian and drone hazard management lies in making sense of the immense volume of data produced by detection systems. From radar returns to thermal imaging and acoustic signatures, the sheer quantity of information can overwhelm even the most sophisticated systems. Sorting through this sea of data to identify actionable insights requires a delicate balance of technology and human expertise.

Radar technologies, such as frequency modulated continuous wave radar and s-band radars, are vital for wide-area detection but often struggle with clutter and time delay issues. Similarly, drone detection systems using radio frequency-based tracking or optical sensors grapple with differentiating between drones and birds — a challenge that complicates real-time decision-making. The integration of these technologies, supported by AI algorithms capable of real-time processing and prioritisation, offers a promising way forward but remains a work in progress.

The limits of current solutions
Commercially available avian detection systems are in use at airports across the world. Several newer experimental systems have shown increased promise, though none fully address the dual challenges of accuracy and integration. For example, avian-specific radar systems excel at detecting bird movements but are limited by the altitudes and ranges, which are necessary to allow time for predictive avoidance. Similarly, drone detection systems equipped with Remote ID capabilities can reliably track uncrewed aerial vehicles but struggle with non-cooperating targets and require further refinement to filter out high false positive rates caused by non-threatening objects.

These limitations highlight a recurring theme in avian and drone hazard management: the gap between theoretical capability and real-world performance. Bridging this gap demands not only technological innovation but also regulatory participation and cross-industry collaboration. Without standardised metrics and guidelines, even the most advanced systems will struggle to gain widespread adoption.

Looking ahead
Future developments should include a focus on refining detection and identification technologies (see and distinguish) and flexibly adapting their operational integration with individual airport operating environments. Key recommendations include:

• 1. Enhanced use of bird data in AI models: There is more than 20 years of data on bird activity that can be used to train and validate predictive models.
• 2. Standardisation: Identifying a means to systematically assess drone risks by the size of individual drones would help to standardise detection thresholds.
• 3. AI-Driven insights: Developing machine learning tools capable of distinguishing between birds and drones in real time could alleviate cognitive load on air traffic controllers.
• 4. Pilot programs: Testing integrated systems in real-world environments, such as the Tāwhaki National Aerospace Centre in New Zealand's South Island, and at major airports could provide valuable feedback and help demonstrate new technology effectiveness under authentic conditions.

Conclusion
The management of hazards presented by birds and drones is a dynamic and ongoing challenge that requires a multidisciplinary approach. From the technical intricacies of radar calibration to the economic pressures imposed by operational disruptions, the path forward is anything but simple. Yet, by leveraging advanced technologies and fostering collaborative innovation, the aviation industry has an opportunity to transform a reactive process into a proactive strategy.

Ultimately, the goal is not merely to detect and respond to hazards but to create an ecosystem where such risks are minimised from the outset. As we continue to navigate the complexities of a crowded and unpredictable airspace, one thing remains clear: success will come not from drowning in data but from learning to swim in it.

Acknowledgements
The author would like to thank Shaun Cole-Baker, Yazan Sinjab, Nathan Pages, and Robert McGivern for their contributions to this piece.

Learn more about Beca’s Defence & National Security capability, including our work in the Aerospace domain here.