Application of a method for the automatic detection and Ground-Based Velocity Track Display (GBVTD) analysis of a tornado crossing the Hong Kong International Airport

A weak tornado with a maximum Doppler velocity shear of about 40 m s− 1 moved across the Hong Kong International Airport (HKIA) during the evening of 20 May 2002. The tornado caused damage equivalent to F0 on the Fujita Scale, based on a damage survey. The Doppler velocity data from the Hong Kong Terminal Doppler Weather Radar (TDWR) are studied using the Ground-Based Velocity Track Display (GBVTD) method of single Doppler analysis. The GBVTD analysis is able to clearly depict the development and decay of the tornado though it appears to underestimate its magnitude. In the pre-tornadic state, the wind field is characterized by inflow toward the center near the ground and upward motion near the center. When the tornado attains its maximum strength, an eye-like structure with a downdraft appears to form in the center. Several minutes later the tornado begins to decay and outflow dominates at low levels. Assuming cyclostrophic balance, the pressure drop 200 m from the center of the tornado at its maximum strength is calculated to be about 6 hPa. To estimate the maximum ground-relative wind speed of the tornado, the TDWR's Doppler velocities are adjusted for the ratio of the sample-volume size of the radar and the radius of the tornado, resulting in a peak wind speed of 28 m s− 1, consistent with the readings from a nearby ground-based anemometers and the F0 damage observed. An automatic tornado detection algorithm based on Doppler velocity difference (delta-V) and temporal and spatial continuity is applied to this event. The locations and the core flow radii of the tornado as determined by the automatic method and by subjective analysis agree closely.

► We analyze the structure and evolution of a tornado crossing the Hong Kong airport using radar and surface meteorological data and GBVTD.
► Inflow towards the center prior to genesis, and a central downdraft at maturity are diagnosed.
► An automatic tornado detection algorithm is evaluated and performs well.
► Radar measured winds are consistent with observed damage.