The rapid expansion of Unmanned Aerial Vehicles (UAVs) and Advanced Air Mobility (AAM) in urban environments poses significant challenges for Unmanned Traffic Management (UTM), particularly in predicting and mitigating safety risks in complex low altitude airspace. UAV operations are subject to dynamic hazards such as dense traffic, infrastructure obstacles, adverse weather, and off-nominal events including system failures. Under such uncertainties, UTM systems must evaluate and compare the risks of alternative flight routes across different operational scenarios, which is critical for both strategic planning and tactical deconfliction. Recent studies have demonstrated the value of dynamic risk mapping for AAM systems. For example, Altun et al.[1] showed that risk maps can support contingency management by enabling safer route selection during emergency conditions. However, existing approaches lack probabilistic frameworks capable of integrating heterogeneous uncertainties while simultaneously comparing multiple routing options. Bayesian Networks (BNs) provide an effective solution by modelling conditional dependencies and supporting probabilistic reasoning under uncertainty. Their ability to combine empirical data with expert knowledge and dynamically update predictions makes them particularly suitable for UTM applications [2]. This study proposes a Bayesian Network-based probabilistic risk assessment framework to evaluate and compare safety risks across alternative UAV routes under varying scenarios. By integrating traffic, system, environmental, and infrastructure factors, the model generates scenario-based risk predictions to support data-driven decision-making. The framework contributes to safer and more adaptive UTM systems through risk-informed routing strategies.