Long-Term Statistical Properties of Extreme Rainfall Data in Japan

In radio meteorology, a field focused on meteorological phenomena that affect radio wave propagation, rainfall, which causes radio wave attenuation, is of significant interest. The modeling of rainfall intensity characteristics is being carried out throughout much of the world, producing a database that is used for estimating the effect of rainfall attenuation on wireless links. Since rainfall characteristics vary greatly from year to year, 10-20 years of accumulated data are required to obtain stochastically stable characteristics. In recent years, global warming has emerged as a serious issue, with global temperatures reportedly increasing by 0.73 degree celsius in the past 100 years; in Japan, the increase has been 1.21degree celsius. If such significant long-term changes have occurred in rainfall characteristics, the rainfall intensity database being used for wireless system design will need to be revised. In this article, we examine rainfall statistics in Japan and assess whether they provide evidence of long-term change. The data include the maximum one-day, 1-h, and 10-min rainfall for each year over the past (approximately) 100 years, as provided by the Japan Meteorological Agency (JMA) on its website. Using a basic confidence interval framework, the statistical properties of the annual maximum rainfall are investigated to clarify, whether the data indicate a long-term trend in rainfall characteristics over the period under the investigation. As a result, this analysis confirmed the existence of a long-term tendency for the three categories of rainfall in Japan to increase by roughly 10% over the examined period of approximately 100 years. In addition, by focusing on statistics on heavy rainfall that occurs once in N years, the study also showed that effective quantitative estimation is possible by applying extreme value statistics.