Rainfall Records & Extreme Value Theory

This week’s horrible events in India, with a one-day rainfall of as much as 37 inches, plus more than 600 dead, is dreadful & dispiriting news. And not to in any way diminish its awfulness, it did get me thinking about rainfall records. How often do such deluges happen? Is it truly a rare event? What is the most rain that has ever fall in one place in one day? Where does more than three feet, which is what fell in one day last week in Mumbai, fit into the picture?
The answer is as shown in the following table (from a truly excellent bathroom book, “Extreme Weather”):

The truly remarkable thing is that the tragic recent downpour in India isn’t even in the top fifteen in the last hundred years. The list includes such otherworldly one-day rainfalls as the more than six feet that hit Reunion Island in 1952.
Remarkable stuff, and a real validation of some of the tenets of extreme value theory. Natural systems are prone to extreme deviations, much more extreme than most of us can imagine — and, sadly, so extreme that such events sometimes take lives.


  1. Cool stats. Most remarkable thing if you look at the world stats is how there are no maximas from 1920’s, 30’s or 40’s. A bunch from 50’s,60’s, and 70’s and then none again from 80’s, 90’s and 00’s. Seems like there’s some periodic weather pattern that causes a wet-dry switch every 30-40 years. Truly extrapolating a lot from 10 data points :).
    So more than extreme value theory, it appears to be an illustration of natural periodicities that are hard to track in human time scales. We feel comfortable with 1 year, maybe 10 year periodicities, but that about seems to be our limit…

  2. Michael Robinson says:

    Pretty interesting top five list. Four rainfalls on the same tropical island, followed by Inner Mongolia (semi-arid inland steppes).
    I’m skeptical about the reliability of meteorological data out of Inner Mongolia at the end of the Cultural Revolution. Just saying.