The transformation towards sustainable urban infrastructures (smart grids, connected mobility, digital water management and energy-efficient buildings) increases the attack surface and amplifies the propagation of failures due to cyber-physical interdependencies. This paper proposes a multivariate statistical modeling approach to quantify cyber risk by incorporating environmental variables (extreme temperature, heavy precipitation, coastal flooding, air quality) as modulators of vulnerability, exposure, and impact severity. The proposal integrates (I) a probabilistic layer for causal and temporal dependencies using dynamic Bayesian networks, (II) a classification and calibration layer with logistic regression to estimate probability of loss events, and (III) a Monte Carlo simulation module to estimate loss distributions and operational metrics (e.g., interruption time). An illustrative case with typical smart city variables is presented and expected results are discussed in terms of increased probability of incidents during extreme environmental events, consistent with evidence on risks in interdependent critical infrastructure. The approach contributes to investment and resilience decisions, aligned with contemporary cyber risk management frameworks.