Comparative study of the relationship between regional climate factors and local weather in Western Pacific countries

摘要

Many studies have reported that climate change affects the local weather. However, there are still considerable uncertainties, since the results depended on the various regional characteristics, time scales, and statistical models. The purpose of this study is to examine how regional climates such as El Nino-Southern oscillation and Indian Ocean Dipole can influence the local weather, and to compare the patterns between some countries in Western Pacific. Monthly data for regional climate and local weather, such as NINO3, DMI (dipole mode index), maximum temperature, and rainfall, were derived from Regional Office of Western Pacific, concerning 21 regions or cities in Mongolia, South Korea, Japan, Papua New Guinea (PNG), Laos, and Cambodia. We defined the four seasons in former three countries, but wet and dry seasons in latter countries. Generalized linear models with interaction between season and other explanatory variables (climate or weather) were used to estimate NINO3 and DMI effects on maximum temperature or rainfall. Also, the moving average lag structures of NINO3 and DMI were applied in the model. The significant NINO3 and DMI effects on rainfall were mostly shown in summer (Mongolia, Korea, and Japan) or all season (Laos and PNG), but the direction varied region to region. With respect to maximum temperature, NINO3 and DMI effects were relatively consistent within the country, but varied on season. For instance, the effects in summer (negative on NINO3, positive on DMI) were in contrast with other seasons in Mongolia, Korea, and Japan, in addition NINO3 effect of PNG was negative in wet season, but positive in dry season. Meanwhile, the lagged effects of NINO3 and DMI on rainfall and temperature were mixed from long and short lags in most countries, whereas only short-term effect was dominant in PNG. Local weather was not only influenced by regional climate factors, but the patterns varied on season, time scale, and locations. These results would be able to support understanding the relationship between weather and health problems.