INCORPORATING INCLEMENT WEATHER IMPACTS ON TRAFFIC ESTIMATION AND PREDICTION

摘要

Traffic simulation models should reflect a wide variety of operating and environmental conditions in orderrnto be useful in estimation and prediction of real world traffic behavior. One area that traffic simulationsrnhave to accurately model is the effect that weather has on driver behavior and traffic flow characteristics.rnRecently, the Northwestern University and the University of Virginia have addressed this deficiency byrninvestigating observed impacts of weather on both the supply and demand of traffic networks. This paperrnhas two research purposes: to present the methodology for incorporating weather impacts into thernDYNASMART-P traffic estimation and prediction tool and to assess the relative accuracy and fidelity ofrnthe developed weather module. The results indicate that the weather adjustment factor module developedrnby the Northwestern University is capable of reducing overestimation of vehicle speeds from 2 mph to 1rnmph for light rain conditions and reducing the absolute error in estimating speeds during heavy rain byrn33% when compared against simulations performed using normal parameters. Additionally, of the threernscenarios tested for two different weather conditions, the weather adjustment factor method consistentlyrnproduced the smallest root mean squared error between simulated and observed speeds.