Office of Operations
21st Century Operations Using 21st Century Technologies

Integrated Modeling for Road Condition Prediction Phase 3 Project Report

Executive Summary

Transportation systems management and operations (TSMO) is at a critical point in its development due increased data availability and analytics. New approaches in road weather management are bringing together meteorology, traffic management, law enforcement, maintenance, and traveler information to support agency decision-making and influence travel behavior. Through these operational efforts and private sector innovations, travelers today have higher expectations for their travel experience. Travelers now participate in generating and validating information, as well as consuming it. This trend will accelerate with deployment of connected vehicle (CV) systems. Within this context, the role of prediction and forecasting will become more important to travelers' transportation and activity choices, as well as to agency decisions in transportation operations. Freight carriers and logistics providers will also benefit in planning routes, times, and delivery schedules.

Based on these opportunities, the Federal Highway Administration (FHWA) Road Weather Management Program (RWMP) has undertaken the investigation, development, and demonstration deployment of an Integrated Modeling for Road Condition Prediction (IMRCP) system. Phase 1 of IMRCP developed the foundational concept of operations and system requirements. The model envisioned is a practical tool that State departments of transportation (DOT) can use to support traveler advisories, as well as maintenance and operational decisions at both strategic and tactical levels. The IMRCP phase 2 work specified, implemented, tested, and evaluated the IMRCP concept in a demonstration deployment. The concept was vetted with a broad stakeholder group and then developed in a straightforward systems engineering process that continued to incorporate stakeholder feedback at key intervals. Working with local and State agencies, the demonstration system was deployed in part of the Kansas City metropolitan area. Performance of IMRCP models and interfaces was evaluated by the research team over a 4-month period of operations in late 2017 with the KC Scout (KC Scout) traffic management center (TMC).

IMRCP phase 3 builds on the prior work to investigate operations applications. The objectives of phase 3 are to:

  • Redeploy the phase 2 model over the same area in Kansas City;
  • Increase the geographical coverage to all Kansas City metropolitan area highways in the KC Scout areas of operation;
  • Add an additional traffic model;
  • Run the system for two winter seasons;
  • Evaluate the system results;
  • Update the system documentation.

The expanded deployment operated for 18 months with enhancements throughout the period. IMRCP system operations and the operations response with KC Scout were independently evaluated by team members not involved in the development of the IMRPC system.

IMRCP provides an interactive map and flexible reporting tools to meet its goal of providing information on predicted road conditions in support of transportation operations. The data that populate these user interface features are kept in a data store that contains both collected data and data generated through forecasting components. IMRCP generates forecasts for traffic and road weather conditions and obtains forecasts from sources outside the system for atmospheric weather, hydrology, work zone plans, and known special events. Current traffic and incident conditions come from the KC Scout advanced transportation management system (ATMS). Current environmental conditions are collected primarily from the National Weather Service (NWS) and other government agencies.

The IMRCP system forecasts traffic and road weather conditions using current and forecasted atmospheric and hydrologic condition data from the data store, collected from the sources previously described. The Traffic Estimation and Prediction System (TrEPS) model estimates and predicts the traffic demand and network states at the zone-to-zone (origin-destination) level. The machine learning-based prediction (MLP) package predicts traffic network conditions based on a given set of system variables, including weather, work zones, incidents, and special events. The Model of the Environment and Temperature of Roads (METRo) model estimates and predicts pavement conditions on roadways within the network of interest.

An evaluation of the IMRCP demonstration deployment was conducted by research team members with the staff of the KC Scout TMC. The evaluation explored what impact IMRCP had on KC Scout operations and assessed if the information was useful to the KC Scout operators/supervisors. The key questions guiding the evaluation data collection and analyses were: (1) Did IMRCP have an operational impact? and (2) Did the users consider IMRCP information useful? The analyses of IMRCP speeds data and operator/supervisor interviews provided insight into the actual and perceived operational impact and usefulness. The evaluation also provided perspectives on the accuracies of the traffic models relative to data provided by the KC Scout traffic management system.

The enhancements made to expand the regional view and traffic modeling capabilities have been integrated successfully into the system. The demonstration deployment provides real-time predictions of traffic and road conditions over the entire Kansas City metro area, which incorporate data on atmospheric and road weather conditions, traffic, incidents, hydrological conditions, work zones, winter maintenance operations, and special events, when such data are available. The predictions are available to system users, operators, and maintainers in user-friendly maps and reports. The underlying IMRCP system provides a scalable framework for deployment in other areas, is extensible to other types and sources of data, and can support additional application-specific user interfaces, as needed.

Lessons learned, deployment experience, potential applications, and conclusions from the IMRCP phase 3 deployment have identified both gaps and barriers to further deployments. Gaps discussed in this report are primarily technical, but barriers may be limited acceptance with operators and maintainers, ease of deployment, or operational benefits. The research team's recommendations for further study are therefore intended to close gaps, reduce barriers, and develop new opportunities for applying predictive capabilities in solving transportation operations challenges. Key recommendations include automating the configuration processes for new deployments, improving data acquisition, adding quality checking to feed the prediction models, tightening the integration of hydrological models, and providing actionable recommendations for management strategies in response to weather and traffic predictions.