Among various weather phenomena, extreme rainfalls have large impacts on our lives and on our societies. However, it is not simple to study extreme rainfalls utilizing satellite data. First of all, definitions of the extreme rainfall differ from place to place over the globe, depending on the regional characteristics of dominant weather systems and associated rainfall. Secondly, the study of “extremes” needs to handle statistically defined rare occasions and always suffers from the scarce sampling of instantaneous satellite observations. Despite of these difficulties, since the Tropical Rainfall Measuring Mission (TRMM) satellite has accumulated three dimensional rainfall data for almost 14 years, it may be the time for us to examine the feasibility of the extreme rainfall study utilizing TRMM data.

In this work, extreme rainfall at each location is defined utilizing TRMM Precipitation Radar data. We first identified the “rain area” by contiguous rain area of the PR2A25 version 6 product. Instantaneous rainfall area size, total rain amount, maximum rain intensity, and maximum rain top height of each rain area were stored for eleven years from 1998 to 2008, in every 2.5degree x 2.5degree longitude and latitude grid. After comparing various parameters, we defined the extreme rain with the 99.9 percentile of the rain areas sorted with the maximum rainfall intensity.

First, an experimental “ground validation” of the above extreme rain is performed against the extreme daily rains obtained from a rain gauge based gridded precipitation data set over the Asian region named “Asian Precipitation - Highly-Resolved Observational Data Integration Towards Evaluation of Water Resources (AFRODITE)” (Yatagai et al. 2008). Although two “extremes” do not consists of directly comparable amounts, they have notable correlation for the latitudes to the north of 5degree N with a correlation coefficient of 0.67. It is suggested that our extreme rain area method well represents the daily rainfall except for the very deep tropical land region, where intense and unorganized afternoon shower may contribute more to the extreme rainfall.

By comparing extreme characteristics over the entire TRMM observation region (36N-36S), it is shown that the extreme rain in the Japan region is characterized with distinguishingly strong intensity and large amount, despite of the relatively low maximum rain top heights. About half of the extreme rain in Japan region is associated with the Baiu (monsoon rain) front, most of the other half consists of rain associated with typhoons or extratropical cyclones, with similar numbers, and some other scattered rain.

In conclusion, we can expect a good feasibility of detecting regional characteristics of the extreme rainfall with the satellite data, provided that we can obtain a long-range homogeneous observation from space.