Road Weather Information System Research Articles

Feasibility of LoRa‐based Infrastructure‐to‐Vehicle (I2V) communication for road weather information systems

SummaryA road‐weather information system (RWIS) is an attractive application in the context of intelligent transportation systems (ITSs), which is used where climatic conditions and meteorological events worsen traffic safety. The primary goal is to significantly reduce potential accidents by sending early warning messages to prompt immediate action. For this purpose, road weather information stations are placed at critical points on the highways. In addition to responsible transportation agencies, enabling direct warning message transfer from roadside infrastructure to vehicles can significantly improve the effectiveness of a typical RWIS application. Dedicated short range communication (DSRC) and cellular vehicle to everything (C‐V2X) technologies are extensively used for data exchange between vehicles (V2V) and between vehicles and road infrastructure (V2I or I2V) under the umbrella term of vehicle‐to‐everything (V2X) communications. Latency and coverage are key performance indicators of the vehicular network that determine how drivers react to risks in adverse road and weather conditions. V2X‐based connectivity mainly aims to offer low latency in message delivery, but it is not possible to obtain a wide‐range coverage due to significant propagation loss in the dedicated 5.9 GHz ITS frequency band. In this study, we presented a long range (LoRa)‐based RWIS application that uses I2 V communication at sub‐GHz bands. Our main contribution is proposing a novel method for RWIS implementations, where the LoRa gateways run in a half‐duplex way to collect information from road weather information stations in the uplink channel and then broadcast warning messages to vehicles in the downlink channel. To validate the applicability of the proposed method, we presented a feasibility study that includes transmission range and latency analysis. Our theoretical calculations' results validate the more extended transmission range and applicability of LoRa connectivity on RWIS applications, especially for common scenarios like black ice where the climate risks do not occur suddenly. We also showed that our methodology would be beneficial due to the ability to offer wide coverage, especially in the areas where cellular base stations have not yet existed or V2X road infrastructure has not yet been established.

Estimating the Expected Change in Safety for a Potential Application of Three Intelligent Transportation System Treatments

This paper documents and illustrates a practice-ready procedure for estimating changes in crash frequency for specific application circumstances for three intelligent transportation system treatments—Closed Circuit Television Cameras (CCTV), Dynamic Message Signs (DMS), and Road Weather Information Systems (RWIS). The procedure will allow an agency to directly evaluate the change in safety that may be associated with a contemplated treatment. In effect, the approach mimics the application of a Crash Modification Function (CMFunction) in that each potential application will, in principle, have its own Crash Modification Factor (CMF). The procedure uses an empirical Bayes framework with safety performance functions (SPFs) for treatment and non-treatment reference sites. The paper also presents those SPFs, which were developed from Pennsylvania freeway data. In principle, this cross-sectional approach can be applied, as it has been, for other safety treatments where safety effects vary with application circumstance and where that variability cannot be captured with conventional before–after studies.

Investigating Road Ice Formation Mechanisms Using Road Weather Information System (RWIS) Observations

Ice formation on roads leads to a higher incidence of accidents and increases winter de-icing/anti-icing costs. This study analyzed 3 years (2019–2021) of Road Weather Information System (RWIS) sub-hourly measurements collected by the Montana Department of Transportation (MDT) to understand the first-order factors of road ice formation and its mechanisms. First, road ice is formed only when the road pavement surface temperature is equal to or below the freezing point (i.e., 32 °F (i.e., 0 °C)), while the corresponding 2 m air temperature could be above 32 °F. Nevertheless, when the road pavement was below 32 °F ice often did not form on the roads. Therefore, one challenge is to know under what conditions road ice forms. Second, the pavement surface temperature is critical for road ice formation. The clear road (i.e., with no ice or snow) surface pavement temperature is generally warmer than the air temperature during both day and night. This feature is different from a natural land surface, where the land skin temperature is lower than the air temperature on cloud-free nights due to radiative cooling. Third, subsurface temperature, measured using a RWIS subsurface sensor below a road surface, did not vary as much as the pavement temperature and, thus, may not be a good index for road ice formation. Fourth, urban heat island effects lead to black ice formation more frequently than roads located in other regions. Fifth, evaporative cooling from the water surface near a road segment further reduces the outlying air temperature, a mechanism that increases heat loss for bridges or lake-side roads in addition to radiative cooling. Additionally, mechanical lifting via mountains and hills is also an efficient mechanism that makes the air condense and, consequently, form ice on the roads. Forecasting road ice formation is in high demand for road safety. These observed features may help to develop a road ice physical model consisting of functions of hyper-local weather conditions, local domain knowledge, the road texture, and geographical environment.

Weather event characterization: a catalyst for improved spatial mapping and benefit quantification in winter road maintenance

Monitoring winter road surface conditions (RSC) is essential for optimizing winter road maintenance operations and ensuring public traffic safety and mobility. However, many existing RSC estimation methods focus primarily on local or site-specific conditions, thus lacking effectiveness in comprehensive spatial mapping under varying winter weather events. Additionally, the potential financial benefits these methods could provide to maintenance authorities remain largely unexplored. This study addresses these gaps by introducing a refined methodology that harnesses the K-Means algorithm to characterize various weather events – a significant stride towards improving the generalizability and accuracy of road surface temperature (RST) estimations. The study also incorporates regression kriging (RK), a renowned geostatistical technique, as an integral component of a system for constructing an extensive spatial map of RSC across large highway networks. This strategy maximizes the use of data from existing road weather information systems (RWIS) to bridge their large spatial gaps. This study further quantifies the potential savings derived from optimally placing RWIS stations and reducing traffic collisions. The efficacy of these methods is validated through a real-world case study on two major interstate highways in Iowa, US, with RST estimation discrepancies as low as 0.619 °C. The findings also indicate potential cost savings by conserving up to 10 RWIS stations, which can be further translated into monetary savings of RWIS capital costs, traffic collisions prevention, enhanced traffic mobility and savings in maintenance material usage.

Strategic Planning for Equitable RWIS Implementation: A Comprehensive Study Incorporating a Multi-variable Semivariogram Model

This paper extends the previously developed method of optimizing Road Weather Information Systems (RWIS) station placement by unveiling a sophisticated multi-variable semivariogram model that concurrently considers multiple vital road weather variables. Previous research primarily centered on single-variable analysis focusing on road surface temperature (RST). The study bridges this oversight by introducing a framework that integrates multiple critical weather variables into the RWIS location allocation framework. This novel approach ensures balanced and equitable RWIS distribution across zones and aligns the network with areas both prone to traffic accidents and areas of high uncertainty. To demonstrate the effectiveness of this refinement, the authors applied the framework to Maine’s existing RWIS network, conducted a gap analysis through varying planning scenarios and generated optimal solutions using a heuristic optimization algorithm. The analysis identified areas that would benefit most from additional RWIS stations and guided optimal resource utilization across different road types and priority locations. A sensitivity analysis was also performed to evaluate the effect of different weightings for weather and traffic factors on the selection of optimal locations. The location solutions generated have been adopted by MaineDOT for future implementations, attesting to the model’s practicality and signifying an important advancement for more effective management of road weather conditions.

Enhancing Autonomous Vehicle Safety in Cold Climates by Using a Road Weather Model: Safely Avoiding Unnecessary Operational Design Domain Exits

<div>This study investigates the use of a road weather model (RWM) as a virtual sensing technique to assist autonomous vehicles (AVs) in driving safely, even in challenging winter weather conditions. In particular, we investigate how the AVs can remain within their operational design domain (ODD) for a greater duration and minimize unnecessary exits. As the road surface temperature (RST) is one of the most critical variables for driving safety in winter weather, we explore the use of the vehicle’s air temperature (AT) sensor as an indicator of RST. Data from both Road Weather Information System (RWIS) stations and vehicles measuring AT and road conditions were used. Results showed that using only the AT sensor as an indicator of RST could result in a high number of false warnings, but the accuracy improved significantly with the use of an RWM to model the RST. ROC-curve analysis resulted in an AUC value of 0.917 with the AT sensor and 0.985 with the RWM, while the true positive rate increased from 67% to 94%. The study also highlights the limitations of relying on dashboard cameras to detect slippery driving conditions, as it may not be accurate enough to distinguish between, for example, wet and icy road conditions. As winter maintenance often prevents slippery roads, the vehicles often measured wet or moist roads, despite RST < 0°C. Our calculations indicate that the vehicle should be able to detect 93% of slippery occasions but the rate of false warnings will be as high as 73%, if using a dashboard camera along with the AT sensor. There are clear benefits of using a RWM to improve road safety and reduce the risk of accidents due to slippery conditions, allowing AVs to safely extend their time within their ODD. The findings of this study provide valuable insights for the development of AVs and their response to slippery road conditions.</div>

Winter Road Surface Condition Recognition Using Semantic Segmentation and the Generative Adversarial Network: A Case Study of Iowa, U.S.A.

Adverse road surface conditions (RSCs) are a significant concern for traffic safety and mobility during the winter season. Dealing with this issue requires the prompt execution of maintenance operations to restore safe driving conditions. The faster maintenance personnel are made aware of the dangerous conditions, the faster they can mobilize to mitigate adverse RSCs. Therefore, the speed at which RSC information is provided is vital for reducing potential road-related incidents and maintaining mobility. Intending to improve the delivery of RSC information, this study proposes developing a deep learning-based method through the usage of road weather information system (RWIS) images to automate the RSC classification process. While the RWIS collects images that give a direct view of the road, the traditional manual way of utilizing these images for RSC monitoring is laborious and time-consuming. To overcome this challenge as well as other limitations associated with previously developed RSC recognition methods, the application of deep neural networks using imagery data is investigated here. More specifically, the application of semantic segmentation (SS) and generative adversarial network (GAN) techniques in automating RSC recognition was adopted in an attempt to predict the locations of drivable areas and indicate the snow hazard level solely based on RWIS images. Case study results from Iowa, U.S.A., show that both SS and GAN techniques perform their tasks with a high degree of accuracy.

Blow-Over Risk Evaluation System for High-Profile Light-Weight Vehicles on Wyoming Highways

Tractor-trailer blow-over crashes caused by strong winds in Wyoming, U.S., can damage roadways, delay goods, and threaten the safety of all members of the traveling public. Identifying roadway sections with high blow-over risk is important for taking appropriate and targeted precautions to avoid these serious problems. This study utilizes a vehicle blow-over model compiled from previous research to quantify incident risk on Wyoming roads and test the appropriateness of existing risk assessments. Data about blow-over crashes from 2012 to 2020 were used to calibrate a new statewide model, and this model was leveraged to observe the evolution of blow-over risk during high wind periods from past blow-over crash events. The results from this model show that blow-over crashes occur in specific zones with elevated risk variables. The model was also especially precise at predicting crash locations within 10 km of nearby weather stations. This computationally efficient model can now be used for real time blow-over risk prediction using road weather information system environmental sensor stations throughout Wyoming. This dynamic risk assessing tool, continuously updated online, can help travelers with high-profile light-weight vehicles through existing traveler information systems, and this model also serves as an addition to the typical wind gust data used by decision-makers managing Wyoming roads.

Winter Road Friction Estimations via Multi-Source Road Weather Data—A Case Study of Alberta, Canada

Road friction has long been recognized as one of the most effective winter road maintenance (WRM) performance measures. It allows WRM personnel to make more informed decisions to improve their services and helps road users make trip-related decisions. In this paper, a machine-learning-based methodological framework was developed to model road friction using inputs from mobile road weather information systems (RWIS) that collect spatially continuous road weather data and road grip. This study also attempts to estimate friction using data from stationary RWIS that are installed far from each other, thereby leaving large areas unmonitored. To fill in the spatial gaps, a kriging interpolator was developed to create a continuous friction map. Slippery road risk levels were classified to provide an overview of road conditions via a risk warning map. The proposed method was evaluated with a selected highway segment in Alberta, Canada. Results show that the models developed herein are highly accurate (93.3%) in estimating friction and identifying dangerous road segments via a color-coded risk map. Given its high performance, the developed model has the potential for large-scale implementation to facilitate more efficient WRM services while also improving the safety and mobility of the traveling public.

Climate change and asphalt binder selection across ontario: A quantitative analysis towards the end of the century

The climate across the world and especially across Canada is becoming warmer. Recent research suggests that, on average, past and projected warming in Canada are about twice those of global values. As part of the Superpave Mix design system, Performance Grade (PG) of the asphalt binder is selected based on historical climatic conditions in relation to the expected in-service temperature spectrum of the pavement surface. Given the effects of climate change, it is important to ascertain how projected ambient conditions may impact pavement surface temperatures and therefore the suitable PG towards more durable and resilient pavement construction in the future. In this study, new regression models, to predict pavement surface temperatures, were developed using air temperatures and latitude of a site. These models were developed to determine the relationships between asphalt pavement surface temperature and ambient weather data from weather stations within the Ontario Ministry of Transportation's Road Weather Information System (RWIS). The developed models were used to predict the pavement temperatures for future climate data from different Global Circulation Models (GCMs) and Representative Concentration Pathways (RCPs), and the results were compared with that of latest Long-Term Pavement Performance (LTPP) models. According to the results of this study, based on the scale of the expected warming, changes in the asphalt binder grades across Ontario are expected in the future. The results of this study also suggest that different types of PGs including 52–34, 58–28, 64–28and 70–22 may need to be adopted by the end-century. This study provides practical recommendations that road authorities and planners may use to adapt paving materials to changing temperature conditions.

Employment of instrumented vehicles to identify real-time snowy weather conditions on freeways using supervised machine learning techniques – A naturalistic driving study

This study introduces the idea of using vehicles as weather sensors to identify real-time weather on freeways in the same context as Road Weather Information System (RWIS) but in a continuous, trajectory-level, and for road segments allocated in the vehicles paths. The study developed a novel approach to detect snowy and clear weather conditions by utilizing real-time data collected from vehicles' external sensors and CANbus. The proposed approach used time series datasets from the SHRP2 Naturalistic Driving Study (NDS), collected during normal driving conditions on freeways. Trips occurring in snowy weather alongside matched trips in clear weather were segmented into time- and distance-based segments such as a one-minute, one-mile, and half a mile. Three assemblies of the input data are considered in the modeling step: data collected from external sensors, CANbus data, and these two data combined. Data analysis was implemented using the Deep Learning Artificial Neural Network, Decision Tree, Random Forest, and Gradient Boosted Trees models. The results indicate that using different segmentation levels provides decent results in detecting snowy weather. The accuracy in estimating the real-time snowy weather was in ranges of 80% to 85%, 71% to 79%, and 73% to 83% for the one-minute, one-mile, and half mile segmentation types, respectively. The GBT model performed the best among all models based on the area under the Receiver Operating Characteristics (ROC) curve, the highest cumulative percentage in estimating the snowy weather using the lower portion of the population, and the highest overall accuracy. Results indicated that an accuracy of 83% in estimating snowy weather conditions could be accomplished using the data collected from external sensors only without accessing CANbus data.

Forecasting short-term road surface temperatures considering forecasting horizon and geographical attributes – an ANN-based approach

The ability to forecast road surface temperature (RST) in advance is an important asset for winter maintenance (WRM) operations. It effectively allows for the reduction of WRM cost through more efficient use of their maintenance resources. However, considering extensive road networks that must be monitored and the extent at which RST varies over time, WRM agencies have constantly been looking for better ways to generate accurate road weather forecasts as they strive to optimize their WRM services and maintain safe travels. To tackle this, this study developed RST forecasting models using an Artificial Neural Network (ANN). RST measurements collected by six selected stationary road weather information systems (RWIS) stations in Alberta, Canada were utilized to validate the feasibility and applicability of the proposed method developed herein. The developed models were found to generate highly accurate results with mean absolute error (MAE) values of 0.64, 1.20, 1.59, 2.16, 2.56, and 3.03 °C for 1-h, 2-h, 3-h, 4-h, 5-h, and 6-h ahead forecasts, respectively. The novelty of this study lies in investigating the probable effect of some external factors on model performance, where it was revealed that forecasting horizon and geographical attributes influenced forecasting accuracies. Upon investigating the hypothesis that locational attributes would affect forecasting accuracies, the results confirmed that accuracy improved with increasing latitude, and decreasing elevations – worthwhile findings that can potentially lead to developing more refined models for generating highly accurate location-specific RST forecasts.

Vehicular Networking Road Weather Information System Tailored for Arctic Winter Conditions

In order to conduct successful long-term service and system architecture development, permanent infrastructures and development environment are essential. For this purpose, FMI is operating a vehicle winter testing track with advanced communication capabilities within ITS-G5 and 5G test network, along with accurate road weather data and services supported by road weather stations, IoT road weather sensor network and on-board weather measurements. The track is in Sodankylä, Northern Finland, where the long arctic winter period of more than half year allows road weather services development in (and for) severe weather conditions. This environment provides appropriate conditions for the development of advanced ITS safety services equally for traditional, autonomous and alternate energy vehicles, tailored road weather services for each special use case and accurate estimation of performance. Not forgetting the energy efficiency of traffic and communication infrastructures themselves, which are critical elements in the development of the future ITS. This paper introduces the test track infrastructures, related research ambitions and future plans.

Development of a Novel Road Weather Information System Location Allocation Model Considering Multiple Road Weather Variables over Space and Time

This study represents an advanced approach to road weather information system (RWIS) network planning. Here, a methodological framework is developed to determine optimal RWIS locations by integrating two analysis domains: space and time. Using a case study, the application of the proposed method is demonstrated using three critical RWIS variables: air temperature, road surface temperature, and dew point temperature. With these three variables, a series of geostatistical semivariogram analyses are performed to construct a single spatiotemporal model named joint semivariogram, which is able to preserve both spatial and temporal aspects. The constructed joint semivariogram is then used to find the optimal RWIS locations for a randomly generated study area using a popular heuristic algorithm—spatial simulated annealing. The proposed method enhances the previously developed RWIS location allocation model by considering both spatial and temporal components of multiple variables. The finding from this analysis reveals that optimal RWIS location strongly depends on the spatiotemporal autocorrelation structure of the variable of interest. Consequently, location solutions generated using the three variables are found to be different from each other. The variation among the RWIS location solutions is then further quantified by developing a spatial similarity index that is used to measure the degree of spatial similarities between different variables. Overall, the findings documented in this study will provide RWIS planners with a more complete and conclusive location allocation strategy and can act as a decision support tool for long-term RWIS network planning.

Hourly road pavement surface temperature forecasting using deep learning models

Road authorities in cold climates regularly apply salt on roads, during winter, to ensure public safety. Pavement surface temperature is a significant parameter affecting snow and ice melting at the onset of a storm. Road temperature below the freezing point of the applied brine causes ice to form on the road surface. Excessive application of salt on the road can have adverse environmental impacts, especially to soil and water quality. Therefore, forecasting pavement temperature can optimize road salt application, while reducing costs, improving public safety, and reducing environmental impacts. This research aims to develop a reliable and accurate pavement surface temperature prediction tool using machine learning techniques. This study employs advanced deep neural network (DNN) learning techniques to predict pavement surface temperature for road salt management purposes. To validate the proposed methodology, this work used hourly solar radiation and air temperature data from Environment Canada, and pavement surface temperature data collected from the Road Weather Information System (RWIS), for sites around the city of Toronto, Ontario. The performance of the proposed DNN model, that integrates a Convolutional Neural Network (CNN) with a Long Short-Term Memory (LSTM), on pavement surface temperature forecasting, was evaluated against four other comparative machine learning methods, LSTM, Convolutional-LSTM (ConvLSTM), Sequence-to-Sequence (Seq2Seq), and Wavelet neural network (Wavenet) models. A dataset of 10,895 samples was collected as an hourly pavement surface temperature record in three timeframes from November 2009 to March 2014 on Highway 401 in southern Ontario, Canada. Experiments included predictions for 1-, 2-, 4- and 6-hours ahead, using as input features air temperature, solar radiation, and present pavement temperature. The proposed pavement temperature forecasting approach resulting from this investigation suggests our new approach is more accurate than previous models. These results reveal that the CNN-LSTM outperforms the four other models and creates predictions closer to the true pavement surface temperature.

Enhancing the Performance of Vehicle Passenger Detection under Adverse Weather Conditions Using Augmented Reality-Based Machine Learning Approach

In response to extreme traffic congestion in metropolitan areas that causes unnecessarily long travel times, high fuel consumption, and excessive greenhouse gas emissions, transportation agencies have implemented various strategies to mitigate traffic congestion. Managed lanes—one of the measures applied worldwide—provide benefits to road users and operating agencies by integrating advanced technologies such as electronic and dynamic tolling systems. However, those agencies already implementing or considering implementing the managed lane strategy are seeking a solution to effectively and properly charge toll rates based on vehicle occupancy and penalize violating vehicles. Vehicle passenger detection systems (VPDSs) have been developed and evaluated worldwide, but limitations still inhibit their full implementation. This study confirms that the performance of the deep learning algorithm, a core VPDS technology, declines under certain adverse weather conditions because of lack of training data sets. The performance of the “you only look once” (YOLOv3) model trained with a normal weather data set decreased by as much as 8.5% when it was tested for adverse weather conditions. In this study, augmented reality (AR) models are developed to enhance the accuracy of vehicle passenger detection (VPDA) by the VPDS by training the algorithm with AR images representing virtual adverse weather conditions. Models trained with AR image sets of various weather categories (fog, rain, and snow) attained VPDA enhanced by up to 7.9%. The final model significantly improves VPDA under adverse weather conditions. The proposed models could be considered for implementation with road weather information systems under adverse weather conditions.

Near Real-time Map Building with Multi-class Image Set Labeling and Classification of Road Conditions Using Convolutional Neural Networks

ABSTRACT Road Weather Information Systems (RWIS) provide real-time weather information at point locations and are often used to produce road weather forecasts and provide input for pavement forecast models. Compared to the prevalant street cameras, however, RWIS are sometimes limited in availability. Thus, extraction of road conditions data by computer vision can provide a complementary observational data source if it can be done quickly and on large scales. In this paper, we leverage state-of-the-art convolutional neural networks (CNN) in labeling images taken by street and highway cameras located across North America. The final training set included 47,000 images labeled with five classes: dry, wet, snow/ice, poor, and offline. The experiments tested different configurations of six CNNs. The EfficientNet-B4 framework was found to be most suitable to this problem, achieving validation accuracy of 90.6%, although EfficientNet-B0 achieved an accuracy of 90.3% with half the execution time. The classified images were then used to construct a map showing real-time road conditions at various camera locations. The proposed approach is presented in three parts: i) application pipeline, ii) description of the deep learning frameworks, iii) the dataset labeling process and the classification metrics.

Real-time computing of pavement conditions in cold regions: A large-scale application with road weather information system

Pavement conditions including pavement temperatures, freezing and thawing depths, and the consequent mechanical performance are the key to the performance and longevity of the pavement. For example, thaw-weakening is a major cause of pavement damage in seasonally-frozen areas covering half of the U.S., leading to huge financial costs for taxpayers. In recent years, the damage has been lessened due to improved practices with Spring Load Restriction (SLR) policies. However, prevalent SLR date prediction methods/tools are still primitive from the perspective of information technology. Such methods/tools are obtained and/or implemented manually with small amounts of data, labor-intensive observations, and/or subjective experience. The paper reports what has been learned from a recent project supported by the Michigan Department of Transporation for the development of a web-based pavement condition prediction and SLR decision support tool: a web-based app called MDOTSLR. MDOTSLR enables access to much more data with little latency and automates data acquisition, processing, and decision making. In this paper, the data innovations and new models that support the functions of the tool will be first introduced. Followed will be the major functions (or services) of the app including software engineering details. Compared with traditional tools without web delivery, this web-based tool automates the acquisition and processing of weather data, GIS data, road weather information system data, and field measurements in real time and thus enables more accurate and convenient SLR predictions. The tool can be easily extended or modified for other road agencies for immediate financial savings in road maintenance and less disturbance to local transportation and economy.

Using Slightly Imbalanced Binary Classification to Predict the Efficiency of Winter Road Maintenance

The prediction of efficiency scores for winter road maintenance (WRM) is a challenging and serious issue in countries with cold climates. While effective and efficient WRM is a key contributor to maximizing road transportation safety and minimizing costs and environmental impacts, it has not yet been included in intelligent prediction methods. Therefore, this study aims to design a WRM efficiency classification prediction model that combines data envelopment analysis and machine learning techniques to improve decision support systems for decision-making units. The proposed methodology consists of six stages and starts with road selection. Real data are obtained by observing road conditions in equal time intervals via road weather information systems, optical sensors, and road-mounted sensors. Then, data preprocessing is performed, and efficiency scores are calculated with the data envelopment analysis method to classify the decision-making units into efficient and inefficient classes. Next, the WRM efficiency classes are considered targets for machine learning classification algorithms, and the dataset is split into training and test datasets. A slightly imbalanced binary classification case is encountered since the distributions of inefficient and efficient classes in the training dataset are unequal, with a low ratio between classes. The proposed methodology includes a comparison of different machine learning classification techniques. The graphical and numerical results indicate that the combination of a support vector machine and genetic algorithm yields the best generalization performance. The results include analyzing the variables that affect the WRM and using efficiency classes to drive future insights to improve the process of decision-making.

Developing Statewide Optimal RWIS Density Guidelines Using Space-Time Semivariogram Models

Preventing weather-related crashes is a significant part of maintaining the safety and mobility of the travelling public during winter months. To help mitigate detrimental effects of winter road conditions, transportation authorities rely on real-time and near-future road weather and surface condition information disseminated by road weather information systems (RWIS) to make more timely and accurate winter road maintenance-related decisions. However, the significant costs of these systems motivate governments to develop a framework for determining a region-specific optimal RWIS density. Building on our previous study to facilitate regional network optimization, this study is aimed at considering the nature of spatiotemporally varying RWIS measurements and integrating larger case studies comprising eight different US states. Space-time semivariogram models were developed to quantify the representativeness of RWIS measurements and examine their effects on regional topography and weather severity for improved generalization. The optimal RWIS density for different topographic and weather severity regions was then determined via one of the most successful combinatorial optimization techniques—particle swarm optimization. The findings of this study revealed a strong dependency of optimal RWIS density on varying environmental characteristics of the region under investigation. It is anticipated that the RWIS density guidelines developed in this study will provide decision makers with a tool they need to help design a long-term RWIS implementation plan.