The impact of cloud hydrometeor advection on NWP model forecasts

Numerical experiments were conducted to investigate the impact of separately advecting hydrometeors on numerical forecast guidance. Numerous 84-hr forecasts were made using the Nonhydrostatic Multiscale Model (NMMB) on the B grid being developed for operational implementation into the North American Mesoscale (NAM) model in 2011 at NCEP/EMC. One set of runs separately advected each of the hydrometeor species, such as cloud liquid water, rain water, and various species of ice from the Ferrier and WSM6 microphysics. Another set of runs advected only the sum of all of the individual hydrometeor species in the form of total condensate, with the assumption that the fraction of each hydrometeor category was unchanged due to advection. Tests included 4-km forecasts over a single domain covering the contiguous United States (CONUS), as well as 4-km forecasts nested within a larger 12-km domain that covered roughly half the area of the large operational NAM domain. Comparisons showed no significant difference in the general spatial and temporal distributions of precipitation except in very localized areas. Objective forecast verification against surface and upper-air observations, as well as the daily CPC precipitation analyses, showed very small differences between forecasts made by advecting only total condensate versus those made by separately advecting various hydrometeor species, whereas the latter used 12-20% more computing time.