Estimating Vehicle Roadside Encroachment Frequencies by Using Accident Prediction Models

The existing data to support the development of roadside encroachment-based accident-prediction models are limited and largely outdated. Under FHWA and TRB sponsorship, several roadside safety projects have attempted to address this issue by proposing rather comprehensive data collection plans and conducting pilot data collection efforts. It is clear from these studies that the required cost for the proposed roadside field data-collection efforts will be very high. Furthermore, the validity of any field-collected roadside encroachment data may be questionable because of the technical difficulty of distinguishing intentional (or controlled) from unintentional (or uncontrolled) encroachments. A method to estimate some of the basic roadside encroachment parameters, including vehicle roadside encroachment frequency and the probability distribution of lateral extent of encroachments, using existing accident-based prediction models is proposed. The method is developed by utilizing the probabilistic relationships between a roadside encroachment event and a run-off-the-road accident event. With some assumptions, the method is capable of providing a wide range of basic encroachment parameters from conventional accident-based prediction models. To illustrate the concept and use of such a method, some basic encroachment parameters are estimated for rural two-lane undivided roads. In addition, the estimated encroachment parameters are compared with those estimated from the existing encroachment data. The illustration indicates that this method can be a viable approach to estimating basic encroachment parameters of interest and, thus, has the potential of reducing the roadside data collection cost.