Tornado Detection with Support Vector Machines

Theodore B. Trafalis1, Huseyin Ince1, Michael B. Richman2

1School of Industrial Engineering, University of Oklahoma, Norman OK 73019, USA
ttrafalis@ou.edu

2School of Meteorology, University of Oklahoma, Norman OK 73019, USA
mrichman@ou.edu

Abstract. The National Weather Service (NWS) Mesocyclone Detection Algorithms (MDA) use empirical rules to process velocity data from the Weather Surveillance Radar 1988 Doppler (WSR-88D). In this study Support Vector Machines (SVM) are applied to mesocyclone detection. Comparison with other classification methods like neural networks and radial basis function networks show that SVM are more effective in mesocyclone/tornado detection.

LNCS 2660, pp. 289-298.

Last modified: