This project is to extend our previous effort on optimizing a region-wide RWIS sensor network and density (Aurora 2010-04) with the objectives of integrating a larger number of case studies and developing better understanding on, and models for, RWIS location and density optimization.

In our previous effort, an innovative optimization framework was introduced to optimize the spatial design of a regional RWIS network by incorporating the ultimate use of RWIS information for spatial inference as well as traffic distribution. The problem was formulated on a basic premise that data from individual RWIS in a region should collectively be used to maximize their overall monitoring quality. The method developed, for the first time, provided decision makers with the freedom to simulate and optimize their RWIS network by balancing the needs of the road users, winter road maintenance requirements, and other respective priorities in locating RWIS stations. Likewise, several RWIS optimal density charts were generated to help determine the optimal number of RWIS stations required to provide adequate coverage in different regions.

Our previous study, however, suffers two limitations. First, only a small number of case studies were conducted, limiting the generalizability of its findings and conclusions. Secondly, the approach developed in the previous study dealt solely with a spatial domain, which does not account the inherent temporal correlation of road weather and surface conditions. Therefore, the primary objective of this follow-on project is to improve the previous method by incorporating a complete set of analysis domains – space and time, and apply it to a larger number of case studies for more general location guidelines.