ABSTRACT
In Sweden, road surface conditions have been assessed with laser-based profilographs since 1987. The Swedish Transport Administration needs a complete description and prediction but not all roads are assessed every year. Therefore, models are needed to forecast and complete the data.
We propose a linear regression model of International Roughness Index (IRI) and rut depth against pavement age on individual 100-meter segments, with a random component pooled for all segments within the same pavement category and county. This method depends heavily on the rate at individual segments. A model that uses explanatory variables like Annual Average Daily Traffic (AADT), road with etc. has not showed better results. Also, the analysis does not show that the rate of change is higher the first year.
We describe available data and corrections for outliers, unregistered maintenance activities and sudden changes. We also show the widths of confidence and prediction intervals for new observations.
