Optimizing the Location of Road Weather Information Systems (rwis) Stations – a Sampling Design Optimization Approach

This study presents an innovative approach to the design of a road weather information monitoring system (RWIS) that optimally combines spatial data on weather-related road surface conditions with data on traffic volume over a state-wide road network. The optimization method minimizes the spatially averaged ordinary kriging variance of hazardous road surface condition (HRSC) frequencies. Since it is desired that an RWIS should also be located at high traffic demand areas, road class data is implemented in the optimization process. Spatial simulated annealing (SSA) is used to search for the optimal RWIS network design by iteratively examining each possible location and accepting designs that ameliorate a weighted sum of average kriging variance and road class detection capability. This novel approach is applied in the optimization of Minnesota RWIS network to illustrate the distinct features of the proposed method, assess the effectiveness of the current location setting, and recommend new additional stations locations. The findings of the study suggest that the method introduced in this study is useful for determining the optimal RWIS station locations and placing a few in addition to the existing stations by incorporating key elements being considered in practice.