Snowfall and rainfall forecasting from weather radar images with artificial neural networks

Discusses problems of the weather forecasting technique with artificial neural networks and describes some solutions. The authors show that the computational time for learning with an acceleration learning algorithm can be reduced by about 10 percent. To overcome the problem of overtraining, a pruning method is introduced and the prediction error is decreased by about 20 percent. Using the data obtained over a winter, the neural weather forecasting technique is more effective than the cross correlation method in producing a substantial reduction of prediction error.