The management of the highway network presents major and complex problems that challenge highway administrators. The problems are compounded by the long-term economic climate, which forced highway administrators to reduce costs. Some have adopted data sampling, others decreased the data collection frequency, and still others reduced the entire pavement management system operation. Unfortunately, these actions were made without detailed evaluation of their impacts on the pavement management decisions. Consequently, time series pavement condition and distress data were obtained from four State Highway Agencies and the Minnesota Road Research project (MnROAD). The data were used to evaluate the impacts of data sampling, pavement analysis length, and data imputation on the variability of the data and the accuracy of pavement management decisions. It is shown that data sampling causes various degrees of errors in pavement management decisions, including the identification of project boundaries, selection of treatment type, and allocation of treatment funds. Furthermore, data sampling yields longer pavement analysis lengths. It is shown that longer analysis lengths do not necessarily improve the analyses. On the contrary, longer analyses lengths have significant adverse impacts on various pavement management decisions. Finally, some of the measured data in the databases were randomly amputated to simulate missing data and imputed using different methods to bridge the gap. It is shown that all methods yield differences between the measured and the imputed data. However, in general, such differences do not significantly impact the pavement management decisions.
Read
