Development of a Probabilistic Subfreezing Road Temperature Nowcast and Forecast Using Machine Learning

Abstract

In this study, a machine learning algorithm for generating a gridded CONUS-wide probabilistic road-temperature forecast is presented. A random forest is used to tie a combination of HRRR model surface variables and information about the geographic location and time of day/year to observed road temperatures. This approach differs from its predecessors in that road temperature is not deterministic (i.e., provides a forecast of a specific road temperature), but rather it is probabilistic, providing a 0-100% probability that the road temperature is subfreezing. This approach can account for the varying controls on road temperature that are not easily known or able to be accounted for in physical models, such as amount of traffic, road composition, and differential shading by surrounding buildings and terrain. The algorithm is trained using road temperature observations from one winter season (Oct-Mar 2016-17) and calibrated/evaluated using observations from the following winter season (Oct-Mar 2017-18). Case-study analyses show the algorithm performs well for various scenarios and captures the temporal and spatial evolution of the probability of subfreezing roads reliably. Statistical evaluation for the predicted probabilities shows good skill as the mean area under the receiver operating characteristics curve is 0.96 and the Brier Skill Score is 0.66 for a 2-hr forecast and only degrades slightly as lead time is increased. Additionally, the algorithm produces well-calibrated probabilities, and consistent discrimination between clearly above-freezing and subfreezing environments.

Analysis of Connected Vehicle Data to Quantify National Mobility Impacts of Winter Storms for Decision Makers and Media Reports

“…Recent studies have leveraged publicly available weather data from the National Oceanic and Atmospheric Administration (NOAA) to demonstrate the impact of precipitation, including rain intensity on interstate traffic speeds [52,53], predicting roadway visibility [54], forecasting road temperatures [55], and evaluating impacts of weather forecast accuracy on driver commutes [56], among others. Figures 5 and 6 represent integrated visualizations using similar High-Resolution Rapid Refresh (HRRR) weather data (precipitation rate and temperature, respectively), as well as instantaneous median speeds recorded on every 0.1 mile section of I-70 for the ten US States it passes through.…”

A Data-Driven Recipe Simulation for Synthetic Rubber Production

“…Especially in studies of time series data analysis, such as the prediction of sea surface temperature, the robustness of the model needs to be considered when learning the ML model [8]. In other studies, temperature prediction algorithms that take advantage of stochastic properties have been introduced because the temperature is not deterministic [9]. Therefore, a new approach is required to compensate for the shortcomings of time-series prediction models and reflect the conditions of various recipes.…”

Automated Model Selection Using Bayesian Optimization and the Asynchronous Successive Halving Algorithm for Predicting Daily Minimum and Maximum Temperatures

“…Many of these approaches have harnessed the power of neural computing techniques, known for their speed and accuracy [10]. Specifically, ML-based approaches to air temperature prediction involve the application of various methods, including Artificial Neural Network (ANN) [21], genetic algorithm-tuned ANN [22], Honey Badger Algorithm-tuned ANN [23], Gene Expression Programming [23], Support Vector Regression [14,17,21,24,25], Multi-Layer Perceptron [1,14], Multi-Variate Adaptive Regression Spline [26], Extreme Learning Machine [26,27], M5 Prime [28], Random Forest [17,26,29,30], Lasso Regression [29], Regression Tree [17], Long Short-Term Memory Network (LSTM) [1,31], GRU-LSTM [32], Convolutional Neural Network (CNN) [29], CNN-LSTM [1,33,34], Simple Recurrent Neural Network with Convolutional Filters [35], and Stochastic Adversarial Video Prediction [35]. Cifuentes et al [18] provided a detailed review of air temperature forecasting approaches using ML techniques.…”