Automatic Detection of Gust Fronts

Hiroshi Uyeda National Severe Storms Laboratory, NOAA, Norman, OK 73069

Search for other papers by Hiroshi Uyeda in
Current site
Google Scholar
PubMed
Close
and
Dusan S. Zrnić National Severe Storms Laboratory, NOAA, Norman, OK 73069

Search for other papers by Dusan S. Zrnić in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

We have developed a procedure that detects and tracks gust fronts automatically. It does not rely on a single method but requires simultaneous operation of two related algorithms. The convergence algorithm measures radial convergence and hence only gusts propagating along radials can be readily detected. The mesocyclone-shear algorithm measures azimuthal shear and is suitable for detecting gusts parallel with radials as well as low-level vortices. Long shear lines that these algorithms detect are classified as gusts whereas symmetric shear features are rejected if their shear and flux or “momentum” are insignificant; otherwise they are classified as low-level vortices. To locate gusts we use second-order polynomials in the range-azimuth plane. It is shown that predicted gust locations from simple linear projections of the least square fitted curves agree very well with actual gust locations.

Abstract

We have developed a procedure that detects and tracks gust fronts automatically. It does not rely on a single method but requires simultaneous operation of two related algorithms. The convergence algorithm measures radial convergence and hence only gusts propagating along radials can be readily detected. The mesocyclone-shear algorithm measures azimuthal shear and is suitable for detecting gusts parallel with radials as well as low-level vortices. Long shear lines that these algorithms detect are classified as gusts whereas symmetric shear features are rejected if their shear and flux or “momentum” are insignificant; otherwise they are classified as low-level vortices. To locate gusts we use second-order polynomials in the range-azimuth plane. It is shown that predicted gust locations from simple linear projections of the least square fitted curves agree very well with actual gust locations.

Save