An experiment on the initial conditions for a mesoscale forecast

Abstract A 6–h forecast starting at midday on a good sea breeze day has been repeated for seven different sets of initial conditions. The initial values for each field could either be given by an 8–h mesoscale forecast valid at 1200GMT or interpolated from a synoptic scale forecast valid at the same time. It was found that changing the initial temperature field had an important effect on the subsequent forecast, but the impact of changing the initial wind field was relatively small. Changing the initial surface pressures had no visible effect on any aspect of the subsequent forecast other than the forecast of surface pressure itself.