United States Department of Transportation
i
Transportation Infrastructure Flooding: Sensing Water Levels and Clearing and Rerouting Traffic out of Danger
-
2017-10-10
-
[PDF-11.41 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Resource Type:
-
Geographical Coverage:
-
Corporate Publisher:
-
NTL Classification:NTL-OPERATIONS AND TRAFFIC CONTROLS-Traffic Flow
-
Abstract:Flooding in urban areas, driven by both precipitation and high tide events, can have a devastating effect on a region’s transportation system and economic viability. In this multi-disciplinary project, the research team developed an approach to help increase the resilience of transportation operations during nuisance flooding and mitigate the danger to drivers and vehicle-related property damage by developing flood predictions, road closure messages, and re-routing. Rainfall data were obtained and used in conjunction with weather forecasts in independent simulation using the Weather Research and Forecasting (WRF) Model to develop rainfall hyetograph forecasts. Overall, WRF simulated rainfall was within the range of uncertainty for a large scale model in capturing the rainfall events with data assimilation and the performance of WRF with data assimilation was similar to previous studies. The sources of bias in WRF rainfall estimates may be due to forecast error, error in the assimilated data, or bias in the initialization or boundary conditions. To predict roadway flooding, the research team adopted the Random Forest data mining approach to identify relationships and patterns between tide levels and rainfall events. Water table level and wind conditions were also used in the model training. The model’s false positive rate was 8.92%; however, most of the false positive predictions were less than 1.5 flood reports. The model had a much higher false negative rate, 31.50%. Importantly, the model did not have any false negative predictions when the true number of reported floods was high. The total daily rainfall value was by far the most important of the variables. The height of the water table did not add appreciable predictive power to the model. The model had some deficiencies, likely due to the limited amount of data and bias in the flood location reporting.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
