Weighted Position and Motion Vector Consensus of Tropical Cyclone Track Prediction in the Western North Pacific

Abstract

Two approaches are developed and tested to improve the unweighted position consensus for 96-, 108-, and 120-h tropical cyclone track guidance in the western North Pacific. A weighted position guidance technique uses a weighting factor for each model that is inversely proportional to how far the 60-, 66-, and 72-h positions of that model are from the corresponding positions of the 11-member position consensus. The weighted position consensus of 96-, 108-, and 120-h track errors for a sample of 24 storms during the 2006 season are consistently smaller than for the unweighted position consensus. In the second approach, a weighted motion vector consensus is developed that uses the same weighting factors as in the weighted position consensus, except that the weights are applied to 12-h motion vectors between 84 and 120 h. This weighted motion vector consensus has substantially smaller errors than the unweighted position consensus, and results in smoother tracks when one or more of the model tracks drops out of the consensus. It is proposed that the weighted motion vector consensus would provide improved guidance for the 96-, 108-, and 120-h tropical cyclone track forecasts.