The rainfall in Matthew was reproduced in a regional modeling simulation using the Weather Research and Forecasting (WRF) model with grid spacing of 2 km. The model captured the general spatial distribution of precipitation and especially that in the intense rainbands. Over eastern North Carolina and southeastern Virginia, over 50% of total modeled precipitation during Matthew occurred during passage of a single rainband that was about 20 km wide in the meridional direction. Several factors helped enhance the intensity of rainfall in the band: 1) Intense low-tropospheric moisture convergence (40–150 g kg-1 hr-1), 2) A thin zonally oriented region (about 2–4 km wide in the north-south direction) located along the north side of the band where motions were conditionally symmetrically unstable up to 700 hPa, and 3) Forcing for upward motion by the near-surface baroclinic zone. Upward motion in the rainband was strongest, resolved as high as 10 m s-1, where conditional symmetric instability was present.
Some prior studies have suggested that cold air damming (CAD) could enhance rainfall in storms similar to Matthew by intensifying the baroclinic zone and providing stronger forcing for upward motion. Because the intense rainfall during Matthew appeared to be related to inland surface frontogenesis, another simulation was run in which the Appalachian Mountains were removed to test the potential role of CAD in enhancing the intensity of rainfall during the storm. With mountains removed, the simulated cyclone moved through the Carolinas faster, resulting in less rainfall. Also, the magnitude of the cross-frontal temperature gradient was reduced by about 25%; however, the intensity of rainfall did not change significantly, suggesting that mesoscale dynamics in the rainband are mostly responsible for the intense rainfall and are not particularly sensitive to the strength of the surface front.