Comparison of western North Pacific Tropical Cyclone aids using storm-related and synoptic parameters.

Abstract

The performance of four western North Pacific Ocean objective aids (One-way influence Tropical Cyclone Model-OTCM; Recurver analog-REC; Total analog-TOTL; and 500 mb steering-CY50) is evaluated using mean forecast error (MFE), systematic error and cross-track (CT) and along-track (AT) statistics. Stratification of these errors by storm-related (latitude, longitude, intensity, 12-h intensity change and size) and synoptic parameters through empirical orthogonal functions (700 mb easterlies/250 mb trough, 700 mb westerlies/250 mb trough, 700 mb easterlies/250 mb ridge, 700 mb westerlies/250 mb ridge) distinguishes between different conditions associated with tropical cyclone motion. The systematic and CT/AT error results reveal distinct biases for each objective aid: 1) OTCM (short-range rightward/slow and long-range leftward/fast); 2) REC (rightward/fast); TOTL (leftward/fast); and 4) CY50 (rightward/slow). The OTCM has the best performance as a result of a small systematic bias. The REC analog has large forecast errors on left-turning and straight-moving storms due to its selection of only recurving analogs. The selection of analogs from the total sample (TOTL) results in errors due to missed recurvature forecasts. Finally, the CY50 has the worst long-range skill because of the lack of the physics of OTCM and climatology influence of the analog techniques.