The National Center for Atmospheric Research Vehicle Data Translator (NCAR VDT) was developed to ingest and process mobile observations obtained from vehicles. The VDT ingests vehicle data and produces outputs of raw, quality checked, and statistically processed data along road segments of user-defined length and time interval. The VDT also includes an inference module, which consists of algorithms used to derive weather conditions by combining vehicle data with traditional weather observations. The algorithms are used to derive precipitation, pavement condition, and visibility along the road segments, with one additional algorithm used to combine the results of these three into a quick assessment of weather conditions along the roadway for the user.

To develop these algorithms, data from the Development Test Environment (DTE09 and DTE10) field projects in Detroit and Minnesota/Nevada Department of Transportation mobile observations were processed by the VDT, along with ancillary weather data, including radar reflectivity, satellite, and surface observations. The data were mined to determine the inputs and structure of the algorithms and tune them for improved performance. Driver reports were used for verification and, where available, surface station observations. This paper presents an overview of the algorithm development and performance, along with the successes and challenges encountered during the process.