Urban congestion refers to the situation where traffic flow in urban areas is impeded due to various factors such as high vehicle density, limited road capacity, incidents, adverse weather, and special events. This paper presents a novel approach to urban congestion modeling through the identification and mitigation of bottlenecks, with a particular focus on understanding the causal relationships between congestion events. The principal contribution of this research is the development of a trajectory metamodel based on semantic events, which effectively captures and highlights the causal correlations between consecutive congestion events. By modeling these correlations, the metamodel enables the anticipation and avoidance of bottlenecks, which are recognized as critical congestion patterns. The findings provide valuable insights for urban planners and traffic management systems in developing more efficient and proactive solutions to urban traffic congestion.