The flame behavior of forward propagating flames through dowel arrays driven by imposed freestream flows was examined. Initial experiments followed the design and implementation of a large, portable, wind tunnel designed for fire spread research. A suspended array of dowels mimicked the small branches and twigs encountered in fine forest fuels. The array was set up at various element to element spacings and subject to a range of flow speeds. Intended as an extension of existing solid fuel array flame spread literature, the results were unexpected. It was found that at certain flow speed-spacing conditions the flame spread did not quicken with higher flow speeds. This regime of flame spread inversion coincided with a change in flame behavior which resulted in fragmented, independent flames, rather than one continuous flame area.  Because the large wind tunnel did not offer fine flow control and was designed for a high turbulence intensity flow, further experiments were conducted in a modified benchtop wind tunnel. The finer resolution flow control allowed for the mapping of the discrete flame behavior regime previously identified in the large wind tunnel. Furthermore, detailed analysis of the flame shape, ignition behavior, spacing, and flow effects was conducted. Lastly, a time scale analysis revealed a Damkholer number-Stanton number relationship which predicted the threshold between flame behavior regimes.  Existing fire spread models predict faster flame spread with increasing flow speeds. In order to incorporate the effects of the changing mixing conditions and flame behaviors, a stagnation point flow model is employed for the heat flux. The model also incorporates a mass transfer B-number which serves as a surrogate for the changes in flow conditions. The B-number analysis of the array experiments yielded an empirical model based on the experiments in the small wind tunnel. The empirical model predicted the 1 cm spaced array ignition times well.