Thanks to Rick Perez, PE for pointing out this interesting research on landscaped medians. Originally, the Washington DOT was concerned about the negative safety implications of planting trees in boulevards – they obstruct sight lines. A summary of the before/after results from the report's abstract:
Total, fatal, fixed object, pedestrian/bicycle, tree, curb/median and U-turn accident rates were analyzed. The percentages of intersection-related, driveway-related, and other accidents were analyzed. Overall, accidents decreased significantly for the treatment locations, but no other types of accidents showed significant changes at either the treatment or the control locations, except for the percentage of other types of accidents at the control locations, which increased significantly.
It appears that the installation of landscaped medians does not have a detrimental effect on safety and may result in an overall decrease in accidents, conclusions consistent with those of the first phase.
Without boldly saying it – the research says crashes decrease on arterials with landscaped medians. I don't think this is surprising since the before conditions and control conditions had no medians (they either had two way left turn lanes or no median/center turn lane). It is noteworthy that the trees didn't seem to cause problems. As usual, we could use more research to confirm the results but this is quite encouraging for us who believe in Context Sensitive Design.