One in five serious or fatal road traffic accidents occur during severe weather conditions. Even with progress in traffic safety, work remains to be done before achieving a globalVision Zero without casualties or serious injuries from road traffic. The vehicle fleet will become increasingly connected and autonomous, generating information for each kilometer traveled, information that can be implemented to enhance road safety. One way of using these data is to monitor tire-road friction (TRF) to improve knowledge of road surface conditions and the interaction between tire and road. Since 2019, the Swedish Transport Administration has obtained connected vehicle (CV) data, sometimes referred to as floating car data (FCD), to follow up on TRF. The focus has been on how CV data can be applied to support and improve winter road maintenance on Sweden’s public road network. Within the Digital Vinter project, CV data has been validated andanalyzed alongside conventional TRF estimation methods and in relation to road weather information systems (RWIS) and mobile reporting of ploughing (MIP).

This thesis presents the results of seven papers, three journal articles, and four peer-reviewed conference papers. Paper A focuses on proof of concept, investigating potential temporal and spatial coverage in relation to annual average daily traffic (AADT). It also shows that the three independent suppliers of CV data display correlations, indicating that they can partly validate each other. Paper F continues this theme, analyzing coverage across different operational areas with varying road types and traffic intensity. It also explores seasonal behavior by comparing the performance of CV data in summer and winter. The results of Paper A and Paper F highlight the importance of understanding data behavior for proper interpretation. Paper C compares two of the CV data suppliers using confusion matrices to evaluate large-scale correlations. For individual one-hour measurements where both suppliers provided TRF estimates on the same 550-meter roadsegments, correlations ranged from 85% to 93% during winter. Paper E analyzes TRF data from CV fleets in relation to RWIS data and MIP actions. The results show that the CV data can capture TRF before, during, and after snowfalls in real time. Paper E also discusses the potential for integrating this information into winter road maintenance decision-support systems through machine learning.

Several field tests were conducted during Digital Vinter. The first was in Björli, Norway, in 2020 (Paper B), where three suppliers each participated with an individual vehicle. TRF was also measured with a RoAR MK6 (ViaTech), which continuously records TRF. Paper G presents results from a similar field test in Kiruna, Sweden, in 2024, where two suppliers participated with two vehicles each. These were analyzed in conjunction with cloud-based fleet data and conventional systems such as ViaFriction and Coralba.

In both Björli and Kiruna, the results showed that all vehicles and systems detected high and low TRF on both homogeneous (proving ground) and inhomogeneous (public roads) surfaces. Papers B and Paper G discuss how conventional continuous systems, such as ViaFriction and RoAR, generally report TRF values 0.05-0.10 lower than CV fleets. At the same time, both CV data and conventional systems display the same relative fluctuations along the road surface. Differences can partly be explained by tire types: connected vehicles used standard winter tires, while ViaFriction uses airplane tires (Trelleborg T520).

Additional field tests were conducted in northern Sweden, including one in Luleå in 2021 (Paper D). Paper D concludes that even with a constant offset between CV data and conventional systems, 80% of the TRF estimates align within a tolerance of 0.05. As discussed in Paper B, one reason conventional systems tend to show lower TRF values is to maintain a safety margin for winter maintenance, since no single tire can represent an entire national fleet. However, if TRF values are set too low, there is a risk of excessive maintenance, leading to higher costs and environmental impacts.