Weather Applications and Products Enabled Through Vehicle Infrastructure Integration (VII)

10. Overview of Research Needs

REQUIREMENTS FOR MAKING EFFECTIVE USE OF VII DATA FOR ENHANCING ROAD WEATHER INFORMATION

VII will support the development of weather-related products for the surface transportation community. However, it is evident that the use of VII-enabled data for product development and enhancement will also be beset with challenges. In order to make effective use of mobile data for weather-related applications, it will be necessary to invest in research to understand issues associated with current and anticipated data elements. This section of the paper discusses a sampling of VII-related research and development topics that are needed to support the development and improvement of weather and road condition related products.

10.1 Probe Message Processes

Data loss and data latency are critical areas of focus in analyzing probe message process alternatives. As discussed in section 4.2, VII probe message processes will have a critical impact on the development of VII applications. One vital aspect of the probe message processes that may have a considerable impact on weather-related applications is the method by which snapshots are generated and stored on each vehicle. Vehicle data most essential to some weather applications and algorithms will be associated with periodic snapshots. Periodic snapshots are considered to be of lower priority as compared to event and start/stop generated snapshots. In the case of a full buffer, periodic snapshots will be deleted in favor of other snapshots. Under some circumstances (e.g. heavy traffic), this characteristic will result in data loss on spatial and temporal scales necessary for some applications. Data latency may also be an issue related to data delivery via the VII network. Under the current VII framework, vehicles will be capable of storing snapshots that contain environmental and road condition data. The possibility of using some snapshots will depend on how often a vehicle comes within communication range of an RSE. To ensure the stability and accuracy of weather-related VII applications, research is required to examine, design, evaluate, and demonstrate the impact of probe message processes on applications under various conditions.

10.2 VII Adoption Rates

The deployment rate of VII technologies will determine how, when, and where applications can be developed and implemented. For most weather-related applications, there will be some minimum number of data points necessary to produce accurate, timely products. Below this threshold, the impact of VII data on weather and road condition algorithms or applications may be low. The deployment of VII will likely take place over a significant period, with initial deployment in urban areas followed by rural areas. Once the VII hardware is deployed, there will be a gradual increase in vehicle data uptake rates as more vehicles equipped with on-board units begin transmitting data to RSEs. Because of the probable variation in data density from region to region, research will be required to understand and document the amount of data that will be required to support various weather applications.

10.3 Data Processing

Appropriate sampling methods need to be defined in order effectively use VII data. A key advantage of VII is the amount of data that will be made available to support weather and road condition applications. However, this can also be seen as a disadvantage. As deployment of roadside and on-board equipment increases, the magnitude of vehicle data will also increase. It is not presently clear exactly how much data would flow through the VII network, but it will be sizeable. This may result in situations where there is excessive data within the domain of interest. Thus, VII research should include examining the use of statistical techniques to process large amounts of data. It may be found that it is possible to translate VII data into points, segments, grids, and profiles without losing information. This type of statistical processing would facilitate the use of large amounts of vehicle information in some algorithms and applications. Many of these issues will need to be addressed as part of the development process for the WDT.

10.4 Data Quality and Accuracy

The range of data quality and accuracy among various vehicles needs to be established to estimate error more effectively. In a recent study, air temperature observations were taken from mobile sensing platforms along a stretch of road west of the Washington D.C. area (Dulles toll road in Virginia). It was found that the observations were affected by sensor placement, traffic congestion, sun angle and the presence of precipitation. The study also noted that vehicles of the same make and model reported different temperatures under identical environmental and roadway conditions. Finally, a small bias was found in the air temperature observations (8). This type of investigation points to the need for additional research on the quality and accuracy of vehicle generated data. The deployment of VII will result in vehicle data elements from various automobile manufacturers, vehicle types, sensor manufacturers, etc; therefore, it is important that research be conducted to evaluate data quality and accuracy issues associated with use of various vehicle data elements.

10.5 Quality Checking

Quality checking tests will be needed to filter anomalous data from faulty vehicle sensors. Stationary weather platforms such as the NWS ASOS stations are remotely monitored. These platforms also have some limited quality control algorithms designed into the system. Should a problem arise that cannot be fixed remotely, a technician can be dispatched to rectify the problem. In the case of vehicle data, it is unlikely that vehicle operators will have the capability to monitor the output for problems. Even when an issue is identified by an operator, it could be days or months before the vehicle is serviced, and the sensor or system linked to the problem returned to a normal operating state. For this reason (and others previously mentioned), the use of data quality checking procedures on VII data will be necessary to ensure the highest quality data possible. Research is needed to explore the types of quality checking procedures that could be implemented. This research may include investigating the use of current quality checking techniques on VII-enabled data, the use of ancillary data for quality checking, and the development of advanced techniques for mobile platforms.

10.6 Data Fusion

Research is required to investigate the most efficient and effective ways to combine data derived from vehicles with commonly used meteorological datasets such as ASOS, radar, numerical model, and satellite data. The utility and value of ingesting mobile data into other weather products should also be examined. In order to create weather-related applications utilizing VII-enabled data, it will be necessary to combine vehicle data with other complementary data sets using data fusion techniques. The temporal and spatial resolution of VII-enabled data will be a great deal higher than the majority of datasets presently available to the meteorological community. These characteristics would contribute to the identification of small-scale weather features, micro-climates, and localized road conditions; however, new data fusion techniques will need to be developed to take full advantage of VII data.

10.7 Numerical Weather Prediction Model Forecasts

The availability of VII-enabled data will enable improvements in the ability of numerical models to forecast changes in the atmosphere, including the atmospheric boundary layer. The ability of models to forecast accurate boundary layer conditions is dependent on four primary factors: the spatial resolution of the model, the effective simulation of dynamics at various scales, the parameterization scheme used to characterize surface and turbulent processes, and the accuracy of the initial atmospheric structure and surface parameters (9). While the capacity of numerical models to forecast surface conditions is dependent on more than simply defining the initial state of the atmosphere, it is clear that accurate characterization of the atmosphere is an important component in the prediction process. Research is needed to explore the impact of VII data on weather model forecasts of the boundary layer, including data assimilation of mobile data, quality and quantity requirements, and the utilization of indirect atmospheric measurements (e.g. wiper state).

10.8 Human Factors

A number of anticipated weather and road product improvements resulting from VII-enabled data are based on indirect measurements of environmental and road conditions. An example would be the use of windshield wiper state to infer the presence or lack of precipitation. The way in which one person operates a vehicle will differ considerably from another. These differences could be attributed to a number of factors such as age, experience, vehicle, etc. Research is needed to investigate how various segments of the population use common vehicle systems (e.g. wipers, lights, brakes, etc.) during normal and adverse weather and road conditions. This information would be incorporated directly into algorithms and applications, and result in more accurate analyses and forecasts of road weather parameters.