Finding optimal routes for networks of harvest site access roads using GIS-based techniques

Many existing forest access roads are not simple independent paths designed to access a single stand; rather they form complex networks of branching roads designed to access multiple stands. Traditional methods of finding optimal locations for these road networks have been almost entirely manual. This study developed and tested three GIS-based road network location strategies. A Monte Carlo approach was used to compare the solutions generated by each of these strategies to optimal solutions created through a complete enumeration approach. This analysis revealed that a solution technique called branch evaluation required only a fraction of the time and produced results almost as efficient as those developed through complete enumeration. Branch evaluation was then used to recreate actual road networks randomly drawn from the database of roads in the Roosevelt and Arapaho National Forests. It was found that branch evaluation was able to recreate routes similar to the actual road networks over 80% of the time. This relatively high matching rate implies that branch evaluation may be a useful tool for performing automated preliminary design of road networks.