予測バイアス除去手法がウェザールーティングの精度におよぼす影響に関する研究

概要

論文の詳細を見る
Numerical Weather Prediction products used in ship's weather routeing contain significant forecast errors due to imperfect initial conditions and or imperfect knowledge of the mechanic governing the evolution of the atmosphere, a part of these errors occurs systematically (called bias). The optimal routes computed using such forecasts show a deviation from the actual optimal route. Because these biases are so systematic, statistical methods can be successfully used to correct for them. In this study the temporal and spatial evolution of forecast errors and their systematic biases, for surface wind and ocean wave fields issued by the Japanese Meteorological Agency are studied. The systematic biases are computed for various training period, and the effect of these removal methods on the accuracy of a minimum time route is investigated by carrying out simulations using a container ship sailing over the Northern Pacific Ocean. The different obtained routes are compared to and scored against the most optimal route obtained relying only forecast analysis. Results show that a dynamic running bias method requiring a relatively short period of forecast data archival achieves relatively good results especially during periods of unstable and or bad weather.
2012-09-25