Based on TREC (Tracking Reflectivity Echo by Correlation) method, selecting single radar CAPPI data, introducing three kinds of artificial neural networks, such as the radial basis function network, generalized regression neural network, and back propagation neural network with wavelet, as well as support vector machine (SVM), one hour forecast experiments of radar reflectivity are made, and the forecast results are compared with those of TREC. Six indexes of hit rate (HR), false alarm rate (FAR), no alarm probability (NAP), critical success index (CSI), correlation coefficient, and root mean square error are applied to evaluate the forecast effectiveness. The results show that when using HR, FAR, NAP, and CSI, the given thresholds play a key role on the evaluation and the smaller thresholds show better results. There is difference between various networks, as well as SVM and TREC. SVM performs better in one hour nowcasting on the whole, compared with TREC on forecasting strong convective development and changes of storms.