Abstract
Over 50 yr have passed since the publication of Sanders' 1955 study, the first quantitative study of the structure and dynamics of a surface cold front. The purpose of this chapter is to reexamine some of the results of that study in light of modern methods of numerical weather prediction and diagnosis. A simulation with a resolution as high as 6-km horizontal grid spacing was performed with the fifth-generation-Pennsylvania State University-National Center for Atmospheric Research (PSU-NCAR) Mesoscale Model (MM5), given initial and lateral boundary conditions from the National Centers for Environmental Precipitation-National Center for Atmospheric Research (NCEP-NCAR) reanalysis project data from 17 to 18 April 1953. The MM5 produced a reasonable simulation af the front, albeit its strength was not as intense and its movement was not as fast as was analyzed by Sanders. The vertical structure of the front differed from that analyzed by Sanders in several significant ways. First, the strongest horizontal temperature gradient associated with the cold front in the simulation occurred above a surface-based inversion, not at the earth's surface. Second, the ascent plume at the leading edge of the front was deeper and more intense than that analyzed by Sanders. The reason was an elevated mixed layer that had moved over the surface cold front in the simulation, allowing a much deeper vertical circulation than was analyzed by Sanders. This structure is similar to that of Australian cold fronts with their deep, well-mixed, prefrontal surface layer. These two differences between the model simulation and the analysis by Sanders may be because upper-air data from Fort Worth, Texas, was unavailable to Sanders. Third, the elevated mixed layer also meant that isentropes along the leading edge of the front extended vertically. Fourth, the field of frontogenesis of the horizontal temperature gradient calculated from the three-dimensional wind differed in that the magnitude of the maximum of the deformation term was larger than the magnitude of the maximum of the tilting term in the simulation, in contrast to Sanders' analysis and other previously published cases. These two discrepancies may be attributable to the limited horizontal resolution of the data that Sanders used in constructing his cross section. Last, a deficiency of the model simulation was that the postfrontal surface superadiabatic layer in the model did not match the observed well-mixed boundary layer. This result raises the question of the origin of the well-mixed postfrontal boundary layer behind cold fronts. To address this question, an additional model simulation without surface fluxes was performed, producing a well-mixed, not superadiabatic, layer. This result suggests that surface fluxes were not necessary for the development of the well-mixed layer, in agreement with previous research. Analysis of this event also amplifies two research themes that Sanders returned to later in his career, First, a prefrontal wind shift occurred in both the observations and model simulation at stations in western Oklahoma. This prefrontal wind shift was caused by a lee cyclone departing the leeward slopes of the Rockies slightly equatorward of the cold front, rather than along the front as was the case farther eastward. Sanders' later research showed how the occurrence of these prefrontal wind shifts leads to the weakening of fronts. Second, this study shows the advantage of using surface potential temperature, rather than surface temperature, for determining the locations of the surface fronts on sloping terrain.
Structure of a Narrow Cold Front in the Boundary Layer: Observations versus Model Simulation