| dc.description.abstract | In our mundane life, traffic is the problem everyone experiences daily. Rain is one of the most prominent factors that influence traffic. A rational understanding of the effect of rainfall on traffic is significant and necessary. Rainfall reduces the visibility on the roads, reduces pavement friction, reduces vehicle performance, reduces roadway speed, increases delay, increases accident risk and causes speed variability. In countries where the drainage system is not proper, water gets coagulated on the street, resulting in waterlogging, lane obstruction and the partial or complete submersion of lanes. In this thesis, the effect of rainfall on roadway traffic is captured using two parameters: visibility in the rainfall and waterlogging on the road. We used an image-based survey to collect the data. 151 people participated in the survey. After preprocessing the data, we created a linear and a quadratic model to understand the effect of rain on roadway traffic, with visibility and depth of the water on the road (waterlogging) as the input parameters and speed of the vehicles as the output parameter. We successfully integrated the linear and the quadratic model into the simulator Simulator Of Urban MObility (SUMO) by modifying the Krauss car-following model to capture the effect of the rainfall on the roadway traffic. The results suggest that both models have low values for all four error metrics: Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Percentage Error (MPE), which means that both models can be used to simulate traffic in rainy weather under the speed limit of 60 kmph. Apart from this, intensity and water depth on the road (waterlogging) parameters are customizable for every road segment. | |
