Contrasting features of hydroclimatic teleconnections and the predictability of seasonal rainfall over east and west Japan

Abstract

Abstract Hydroclimatic teleconnections between global sea surface temperature (SST) anomaly fields and monthly rainfall over east and west Japan (divided along 138° E longitude) are identified for summer (June–August) and winter (December–February) using the concept of global climate pattern (GCP). The analysis indicates that the hydroclimatic teleconnections over both regions vary at both intra‐ and inter‐seasonal time scales. In addition, the teleconnections over the two regions have differing origins. The teleconnection features associated with rainfall anomalies over west Japan have origins in the tropical Pacific and Indian oceans, whereas those over east Japan are associated with high‐latitude SST anomalies. The early summer (winter) rainfall over west Japan is linked to the El Niño Modoki (La Niña Modoki) phenomena, whereas the early summer and winter rainfall anomalies over east Japan are associated with the SST anomaly over the eastern subtropical Pacific and South Pacific oceans, respectively. Having identified the teleconnections, prediction model approaches—a machine‐learning approach, namely support vector regression (SVR), and a hybrid graphical modelling/C‐Vine copula (GM‐Copula)—were developed to forecast the rainfall over both east and west Japan. The predictors were derived from the monthly SST anomalies at different lags (1–6 months), and whereas the hidden, nonlinear relationship was well captured by the SVR approach, the complex association was decidedly better captured by the GM‐Copula approach. Hence, it is recommended for forecasting the rainfall over east and west Japan.

Keywords

• global climate pattern
• hydroclimatic teleconnection
• Japan
• machine learning
• rainfall prediction
• sea surface temperature