11 Identification and tracking of convective cells in 3D reflectivity mosaic of KMA weather radar network

Monday, 26 September 2011
Grand Ballroom (William Penn Hotel)
Sung-Hwa Jung, Kyungpook National University, Daegu, Korea, Republic of (South); and J. H. Lee, G. W. Lee, H. W. Kim, and B. Kuk

Handout (2.0 MB)

Automatic algorithm for identifying and tracking the convective storm cells in 3-dimensional reflectivity mosaic (3D MOSAIC) from KMA weather radar network have been developed. The inherent limitation of single radar such as the small coverage, cone of silence, and mis-detection of precipitation echoes at lower elevation due to earth curvature can be minimized with the 3D MOSAIC instead of a single radar data. However, the discrepancy of radar reflectivity due to calibration bias, different attenuation, and different beam blockage makes difficult the efficient use of 3D mosaic. In order to make an unified 3D MOSAIC, the calibration biases of KMA radars were derived by using a disdrometer as an absolute reference and then by applying intercomparision among different radars. The effect of beam blocking is reduced by simulation with the digital elevation model with a horizontal resolution of 3sec. The identification algorithm distinguishes an individual convective storm cell using reflectivity and volume thresholds similar to TITAN, SCIT, and E-TITAN. The tracking algorithm utilizes fuzzy logic with four membership functions and their weights. The membership functions and weights of the fuzzy logic were objectively determined using the statistical characteristics of the identified storms. The four properties of speed, area change ratio, difference of mean reflectivity, and axis transformation ratio are used to determine membership functions. Each membership function and their weight are derived from manually matched storm pairs in two consecutive time. The algorithms have been verified for several cases of convective storms in summer season. The algorithms have properly identified storm cells and chased successively comparing with manually matched storm pairs. The developed algorithms in this study may provide useful short-term forecasting or nowcasting capability of convective storm cells and provide the statistical characteristics of severe weather.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner