Abstract
An algorithm to detect frontal zones in time–height cross sections of horizontal wind from wind profiler measurements is described. The algorithm works by identifying regions with 1) a strong horizontal temperature gradient, estimated by using a quasigeostrophic thermal wind retrieval, 2) a strong temporal increase in the signal-to-noise ratio at a given range gate, and/or 3) a strong temporal shift in the horizontal winds at a given range gate. The type (e.g., cold or warm) of front is determined by examining the advection field and the characteristics of the boundary. Most weight is given to the horizontal temperature gradient component of the algorithm.
A springtime frontal system and an associated baroclinic wave over South Australia are examined using both routine synoptic observations and analyses as well as data from the profiler. Synoptic observations depict a prefrontal trough and two cold fronts at the surface and a deep trough in upper levels. The tropopause is identified at ∼6 km in one sounding. The algorithm successfully identifies the one main cold front and the lowered tropopause in the polar air. There are also hints of a prefrontal trough and a descending tropopause with the onset of the main cold front. After the passage of the upper trough, the ascending tropopause and the so-called jet front or trailing front are also identified by the algorithm. The latter represents the passage of the upper-level baroclinic wave and the reappearance of a strong jet stream.
Other regions are spuriously identified as fronts. These regions could be the reflection of some short-term meteorological phenomena, such as gravity waves; deviations from the assumed quasi-geostrophy; or simply reflections of noise in the analysis. An examination of the effect of random measurement uncertainties on the frontal analysis gives an estimate of error of around 2 K (100 km)−1 in the horizontal temperature gradient calculations for typical wind errors. The errors on the retrieved advection vary, depending on the wind speed, but are around 25 K day−1 for a ∼20 m s−1 wind speed. These values are typical of the noise in those fields, suggesting that the spuriously defined fronts likely reflect uncertainties in the data rather than actual meteorological phenomena.
Corresponding author address: Dr. Christopher Lucas, Department of Physics and Mathematical Physics, University of Adelaide, Adelaide 5005, Australia.
