Surface Transportation Weather

Background

RAL is a key contributor to the research and development of the weather component of the Federal Highway Administration’s wireless Connected Vehicle program. RAL also continues to support the adoption of the winter Maintenance Decision Support System (MDSS) technology across the nation. 

In the coming year, RAL will continue to expand efforts by developing transportation decision support systems focusing on traffic, incident, and emergency management and maintenance beyond snow and ice control by seamlessly blending the strategic prediction components of the system with tactical short-term weather and road condition technologies.

FY2010 Research & Development:

Vehicle Data Translator

Figure 1. Data collection routes shown in green.
 Figure 1. Data collection routes shown in green.

The Connected Vehicle program has three main goals, namely to increase safety, mobility, and environmental efficiency. This program will ultimately provide vehicle probe data (including weather data) from millions of vehicles that will be available to the weather community to support the diagnosis and short-term prediction of weather and road conditions. With funding and support from the USDOT Research and Innovative Technology Administration (RITA) and direction from the Federal Highway Administration’s (FHWA) Road Weather Management Program, RAL conducted research in FY11 to continue developing the prototype Vehicle Data Translator (VDT). The VDT incorporates vehicle-based measurements of the road and surrounding atmosphere with other, more traditional weather data sources, and creates road and atmospheric hazard products for a variety of users. Analysis in FY11 focused on three key areas: obtaining mobile data from State Department of Transportation (DOT) partners; developing the third version of the VDT; and publishing prior research in peer-reviewed journals.

RAL partnered with the Minnesota and Nevada DOTs to obtain mobile data. In Minnesota, data from the Controller Access Network Bus (CANBus; e.g., lights and wipers) as well as external data (e.g., pavement temperature) will be collected from 80 snowplows and supervisor trucks along the routes shown in Figure 1. In Nevada, similar data will be obtained along I-80.

RAL also refined the VDT in three stages, as shown in Figure 2. The initial stage of the VDT ingests mobile data. If the data are already pre-processed in some way, such as by the Clarus System, then the VDT can simply read the metadata and data from the Clarus output. However, the VDT also has routines to directly ingest mobile data from the CANBus or a data collection and forwarding facility, parse them, and then sort them by time, road segments, and grid cells (the road segments and grid cells are user-defined via configuration files). This stage also reads any extra data sent from the vehicle that originates outside of the CANBus, such as readings from add-on sensors. These data are then passed through a QC Module that tags data that contain invalid geospatial or temporal information (e.g., latitude values greater than 90°N or time of day greater than 23:59:59). All data are passed through the Output Data Handler, which outputs the “parsed mobile data” for use in applications, and also for use in Stage II of the VDT. One such application has already been created, and it can display the data from this stage on a Google Map.

Stage II analyses provide the road segment data using QC’d mobile data. The QC Module examines individual mobile data (CANBus and some add-on sensors) and flags each data point for the relevant QC tests that are listed in Table 3-1 and outlined in Section 6. Ancillary data, such as Clarus surface station data and radar data, are also ingested by the Ancillary Data Ingesters, which perform the same functions as the Stage I Mobile Data Ingestors module (except in this case for ancillary data), including time stamping and geolocating. These ancillary data are then used in some of the QC processes, but they are not QC’ed themselves; however, the ancillary data used in the VDT 3.0 is all QC’d by other means before being incorporated into the VDT. All data are passed through to the Statistics component, where the mobile data that pass QC are used to compute road segment statistics. Examples of these data would include the mean air temperature over an individual road segment for a given time step, or the percentage of windshield wipers activated over an individual road segment for a given time step. All mobile data with QC flags and the statistical data for the “Basic road segment” data are output from Stage II.

Figure 3. Spanish MDSS Display.
 Figure 3. Spanish MDSS Display.

Stage III analyses provide additional value-adds for mobile data. In the Inference Module, fuzzy logic algorithms, decision trees, and other data mining procedures are used to produce the “Advanced road segment” data. Examples of these include combining mobile data with radar, satellite, and fixed surface station data to compute a derived ‘road precipitation’ product over an individual road segment for a given time step. These data are then run through a QC Module that assigns a confidence value to the “Advanced road segment” data assessments.

Maintenance Decision Support System (MDSS)

Since 1999, RAL has led a team of national laboratories in the development of the Federal prototype winter Maintenance Decision Support System (MDSS), a unique decision support system that provides real-time snow and ice control guidance (e.g., treatment times, chemical choices, rates, and locations) for user-defined roadway segments.

Figure 4. Recent changes in the road temperature forecast system led to improved accuracy.
 Figure 4. Recent changes in the road temperature forecast system led to improved accuracy.

In FY11, NCAR completed development for a prototype Spanish MDSS for roads near Madrid, Spain (Figure 3). The Spanish version maintains functionality of the Federal Prototype, and includes new research aimed for the changes in Spanish maintenance practices.  The road temperature forecast, as generated by the Model of the Environment and Temperature of Roads (METRo), was altered to include post-processing. This led to a noticeable improvement in the road temperature forecast (e.g., Figure 4).

Other highlights included continuing to work with Denver International Airport and supporting the FHWA’s annual Road Weather Management Meeting.