The impact of mesoscale data on the simulation of a mesoscale convective weather system

The objectives are: (1) to demonstrate the sensitivity of a numerical simulation of a mesoscale convective weather system to the initial conditions; and (2) to provide further evidence of the need for a high resolution observing system that is compatible with numerical-model initial data requirements. To this end, a series of nine numerical-model sensitivity experiments were conducted in which one or more variables from one or more observations (soundings) were omitted and/or adjusted in the model initialization. Two types of surroundings were available for manipulation in the sensitivity experiments. Specifically, for the particular event being simulated (the 1977 Johnstown flash flood), an independent mesoanalysis was available from Bosart and Sanders (1981). In his analysis, Bosart produced a fine-resolution three-dimensional array of data from which soundings that helped to define mesoscale features could be extracted. These 'nonconventional' soundings were added to the conventional sounding data routinely available for model initialization.