Applying a Multi-dimensional Time-Series Similarity Method to Typhoon-track Prediction

A tropical cyclone is one of the most threatening natural phenomena and can result in great human and economic loss. To reduce the damage and protect people's lives, it is becoming increasingly important to predict the movement or track of a typhoon. Although there are several methods of predicting a typhoon track, the results are not sufficiently accurate to utilize when a typhoon is threatening a country or area. To reduce the prediction error, in this paper a multi-dimensional time series similarity method called Modified A-LTK, Approximation with use of Local features at Thinned-out Keypoints, is applied to the prediction. Our preliminary evaluation indicates that the error between the original data and the predicted data was reduced using Modified A-LTK compared with other existing methods such as DTW and AMSS.