Abstract
Harmful Algal Blooms (HABs) have been observed in all 50 states in the U.S., ranging from large freshwater lakes, such as the Great Lakes, to smaller inland lakes, rivers, and reservoirs, as well as marine coastal areas and estuaries. In 2014, a HAB on Lake Erie containing microcystin (a liver toxin) contaminated the municipal water supply in Toledo, Ohio, providing non-potable water to 400,000 people. Studying HABs is complicated as different cyanobacteria produce a range of toxins that impact human health, such as microcystins, saxitoxin, anatoxin-a, and cylindrospermopsin. HABs may be increasing in prevalence with rising temperatures and higher nutrient runoff. Consequently, new tools and technology are needed to rapidly detect, characterize, and respond to HABs that threaten our water security. A framework is needed to understand cyber threats to new and existing technologies that monitor and forecast our water quality. To properly detect, assess, and mitigate security threats on water infrastructure, it is necessary to envision water security from the perspective of a cyber-physical system (CPS). In doing so, we can evaluate risks and research needs for cyber-attacks on HAB-monitoring networks including data injection attacks, automated system hijacking attacks, node forgery attacks, and attacks on learning algorithms. Herein, we provide perspectives on the research needed to understand both the threats posed by HABs and the coupled cyber threats to water security in the context of HABs.
Introduction
Harmful Algal Blooms (HABs)—explosive growths of cyanobacteria that produce potent liver and nerve toxins—represent a significant and growing threat to global water security. Driven by rising temperatures and increased nutrient runoff, these blooms are no longer isolated incidents but recurring crises. The 2014 event in Lake Erie, which left 400,000 residents of Toledo, Ohio, without safe drinking water due to microcystin contamination, serves as a stark reminder of the vulnerability of our municipal water systems.
To combat this, environmental management is shifting toward high-tech, automated monitoring solutions. These include unmanned aerial vehicles (UAVs), robotic water samplers, and networked Internet of Things (IoT) sensor arrays. However, this technological shift introduces a new dimension of risk: cyberbiosecurity. By integrating digital controls with biological monitoring, we transform water infrastructure into a Cyber-Physical System (CPS), creating digital vulnerabilities that could be exploited to mask or exacerbate biological threats.
The Cyber-Physical System (CPS) for Water
Viewing water security through the lens of a CPS requires understanding how the physical realm (the water bodies and sensors) interacts with the cyber realm (data transmission, cloud processing, and control algorithms). Modern monitoring involves a multi-scale approach, ranging from satellite imagery to localized robotic deployments.
As shown in the figure above, the integration of satellite data with surface-level autonomous vehicles and aerial drones provides a comprehensive view of bloom dynamics. However, each node in this network—from the satellite downlink to the individual catamaran drone—represents a potential entry point for cyber interference. If the cyber layer is compromised, the physical biological threat (the toxins) can be intentionally ignored or mismanaged by the automated systems designed to protect us.
Threat Modeling in HAB Monitoring
The transition to a CPS framework allows for the identification of specific cyber-attack vectors tailored to HAB monitoring networks:
- Data Injection Attacks: Adversaries can inject faulty data into the system to make high toxin levels appear safe. This "stealthy" attack could prevent water treatment plants from activating necessary filtration protocols, leading to public exposure.
- Automated System Hijacking: Attackers could take control of drones or robotic samplers, derailing their monitoring missions or physically damaging the equipment.
- Node Forgery: By impersonating legitimate sensors, malicious actors can feed fabricated data into the network. Low-cost, small-form-factor sensors are particularly vulnerable to this type of forgery.
- Learning Algorithm Attacks: As we rely more on machine learning to forecast blooms, these models become targets. Adversarial attacks can exploit vulnerabilities in algorithms to produce incorrect forecasts, leading to resource misallocation.
Emerging Insights: Aerosolization and Stealth
A critical and often overlooked risk is the aerosolization of HAB toxins. Research indicates that these toxins can be transported at least 30 km inland through the air, significantly increasing the "attack surface" for biological exposure. This makes accurate, uncompromised monitoring even more vital, as the threat extends beyond the immediate water source.
Furthermore, the paper highlights that attacks on water systems can be conducted with high levels of stealth. An attacker does not need to physically poison a reservoir; they only need to poison the data stream. By bypassing standard state estimators, they can ensure that the system reports "normal" conditions while a toxic bloom is actively entering the municipal supply.
Conclusion and Future Directions
Securing our freshwater resources requires a "convergent paradigm" that combines ecology, atmospheric science, chemistry, and electrical engineering. While identifying these threats is the first step, future research must focus on developing "HAB-centric automated systems" designed with security as a primary principle. This includes implementing device fingerprinting to prevent node forgery and utilizing adversarial machine learning to harden forecasting models. As water infrastructure becomes increasingly interconnected with other city systems, establishing robust cyberbiosecurity standards for environmental monitoring hardware is no longer optional—it is a public health necessity.
Summary updated