The objectives for the work included use of SCOS97 data and MM5 simulations to understand meteorological factors in the formation of high ozone concentrations during 4-7 August 1997. Meteorological data for the case study included: (a) observations at 110 SCOS97 surface sites and (b) upper air measurements from 12 rawinsonde and 26 RWP/RASS profilers. MM5 con-tained the PSU Marine Boundary Layer Initialization (MBLI) scheme, quadruple nested grids (horizontal resolutions of 135, 45, 15, and 5 km), 30 vertical layers, minimum sigma level of 14 m, USGS global land-use, GDAS global gridded model analyses and SSTs, analysis nudging, observational nudging, force-restore surface temperature, 1.5 order TKE, one-way continuous nesting, and a MAPS statistical evaluation. Analysis showed the ozone episode resulting from a unique combination of large-scale upper level synoptic forcings that included a weak local coast-al 700 mb anticyclone. Its movement around SoCAB rotated the upper level synoptic back-ground flow from its normal westerly onshore direction to a less common offshore easterly flow during the nighttime period preceding the episode. The resulting easterly upper level synoptic background winds influenced surface flow directions at inland sites, so that a surface frontal con-vergence zone resulted where the easterly flow met the westerly onshore sea breeze flow. The maximum inland penetration of the convergence zone was about to the San Gabriel Mountain peaks, the location of daytime-maximum ozone-episode concentrations. The MM5 simulations reproduced the main qualitative features of the evolution of the diurnal sea breeze cycle in the SoCAB with good accuracy. The position of the sea breeze front during its daytime inland pen-etration and nighttime retreat could be determined from simulated wind fields. The accuracy of predicted MM5 surface winds and temperatures over SoCAB were improved by modification of its deep-soil temperature, interpolation of predicted temperatures and winds to SCOS97 obser-vational levels, use of updated urban land-use patterns, use of corrected input values for ocean and urban surface roughness parameter values, analysis nudging, and 4DDA.