Satori Defense predicts Houston rail crossing blockages using a time-series model trained on 14+ months of real sensor data from TRAINFO — the only real-time monitoring network on Union Pacific lines in the city.
Every minute, Satori polls the TRAINFO ArcGIS FeatureServer for status changes across all 57 monitored crossings. State transitions — when a crossing goes from clear to blocked and back — are recorded with precise timestamps and durations in UTC. After de-biasing for stale events (blockages exceeding 12 hours are excluded), the clean dataset contains thousands of confirmed blockage events spanning more than a year.
From this history, Satori computes a 24-element probability vector for each crossing: the likelihood of a blockage occurring during each hour of the day, calculated separately for weekdays vs. weekends, and weighted toward recent months to capture seasonal train schedule changes.
The base heatmap probability is multiplied by an Environmental Risk Multiplier — a compound score derived from three live data sources:
When environmental factors compound — for example, a rain event coinciding with the 10 AM peak train window — the system flags elevated risk zones automatically.
Across the 57 monitored crossings, Satori confirms predictions against live TRAINFO readings. For blockages exceeding 15 minutes, accuracy exceeds 70%. High-volume crossings like Cavalcade Street and Hogan Street exceed 80% accuracy due to their consistent train scheduling. False positives — predicted but not observed — account for under 20% of predictions.
Marine drayage operators use the heatmap to schedule port pickup windows around high-risk crossing hours, eliminating the “surprise block” that strands trucks for 30+ minutes. EMS agencies use crossing status for real-time rerouting. Insurance underwriters use the historical blockage record for risk modeling at specific crossing locations.
Schedule a deployment consultation. We move at mission speed. Operational within 48 hours for most environments.