Search

Convective-stratiform rain classification is essential because these rainfall types have different physical characteristics. This work evaluates the use of Artificial Neural Network (ANN)-backpropagation to classify rainfall observed by Micro Rain Radar (MRR) at Serpong, Indonesia. Classification is based on radar reflectivity (Z), Doppler velocity, and Liquid Water Content (LWC). During April 14, 2016, and May 4, 2017, rain events were used as training and testing data, respectively. The best ANN-backpropagation architecture for rain classification is a 3-6-1 architecture (input layer-hidden layer-output layer) with a learning rate of 0.6 for convective rain and 0.9 for stratiform rain, respectively. The accuracy of convective and stratiform rain classification obtaining is 86.9565% and 100.0000%. Good accuracy was also found when the result was validated by the BB (Bright Band) method. Thus, the ANN- backpropagation method can classify the MRR data accurately. [ABSTRACT FROM AUTHOR]