Development of a Physical Security Risk Matrix Integrating Heterogeneous Sensor Data and Spatiotemporal Context: Expert-Derived Parameters and Scenario-Based Validation

Physical security sites combine video analytics, access control, and intrusion sensors, yet each subsystem issues alarms per device and per event, leaving the integration of heterogeneous evidence to operator expertise, so risk judgments become inconsistent and slow. Existing frameworks are oriented toward periodic diagnosis and treat spatiotemporal context as site-specific rules or post-hoc coefficients. This study asks whether a matrix that fuses heterogeneous sensor data into a five-level risk grade, defining context as independent indicators and parameterizing threat scores by facility type, converges with independent expert judgment. Six key risk indicators, covering abnormal behavior, spatial and temporal context, access-control anomaly, sensor evidence, and event correlation, were combined in a normalized weighted sum. Twenty experts completed analytic hierarchy process pairwise comparisons, sub-indicator ratings, and assessments of seventeen scenarios across five demonstration facilities. Geometric-mean aggregation produced a consistent group weight set (consistency ratio 0.012) in which temporal context and video-based abnormal behavior ranked highest, and threat behavior profiles diverged into open-access and access-controlled clusters (cross-cluster correlation 0.29). With equal-interval thresholds the matrix reproduced expert judgments within one adjacent level in 88% of scenarios, with no disagreement in the high-risk range, whereas exact agreement was 29%. Calibrating the two upper thresholds by quantile matching raised exact agreement to 47% in-sample and 35% under leave-one-out cross-validation, and the group weights were robust to consistency screening. The result is an independent judgment layer that converts device-level detections into a decomposable, context-adjusted risk grade with expert-derived initial parameters that can be recalibrated with operational data.