Influence of spatial and seasonal asymmetries on long-range tropical cyclone prediction in the western North Pacific

Abstract

The seasonal predictability of tropical cyclones (TC) in the western North Pacific (WNP) reported in previous studies are mainly based under the general consideration that the WNP is homogeneous in terms of its spatial and temporal characteristics. Here we present evidence that the western (Domain 1) and eastern (Domain 2) parts of the WNP exhibit spatial and seasonal asymmetric response to large-scale environments (e.g., asymmetrical sea surface temperature anomalies distribution) leading to distinct spatial and seasonal TC variability in the said domains. Exploring such asymmetries, we propose an alternative approach on the long-range predictability of TC genesis frequency in the WNP during its active TC season (i.e., June-November, JJASON) by separately predicting the TC genesis frequency in two domains (i.e., Domains 1 and 2) in two distinct seasons (i.e., June-August and September-October), respectively. Using a number of climate indices as predictors in different lead times, our regression-based models present its best significant seasonal predictability of TC genesis frequency during JJASON (i.e., r = 0.80, p < 0.01) that essentially captures the spatial and seasonal asymmetry in the WNP. It is expected that this study provides valuable insights on the long-range and more localized TC prediction in support of disaster risk reduction in the WNP region.