The Statistical Myth of ‘Once-in-a-1000-Year’ Weather Events

Written by Dr. Matthew Wielicki on September 20, 2024. Posted in Current News

When the media claims we’re experiencing a “once in 500-year” or “once in 1,000-year” weather event, they’re missing a fundamental point about how data works

And as an Earth Scientist, this is something I’m acutely aware of. In geology, we study the Earth’s long history through rock formations, sediment layers, and fossil records, which help us track major climatic trends and shifts over Earth’s history.

We can read signs of past floods, droughts, and shifts in temperature, but there’s a crucial limitation here: geological proxies don’t capture daily weather extremes.

You might find evidence of sustained climatic conditions that result in long-term sediment buildup or erosion.

For instance, we can see signs of long-term droughts or periods of significant rainfall across millennia, but you’re not going to find a fossil or rock layer that tells you, “Oh, it rained 10 inches in 24 hours on this particular day 2,000 years ago.”

“That’s just not how proxies work. Proxies give us broad trends over long periods of time, not the kind of hyper-detailed weather data required to make statements about rare events like a “1,000-year storm.”

The truth is, these assertions are built on shaky statistical ground, and there’s no real way we can be certain about the frequency of these events given how little data we actually have.

The Statistical Absurdity of Rare Event Claims

When we talk about events with such extreme rarity, such as a “once in 1000-year” flood, we’re referring to statistical probability based on a distribution of observed events over time.

The tail end of any statistical distribution, especially one that measures the frequency of rare, extreme events, is always the hardest to fill in.

Simply put, the more extreme the event, the less data we have to make reliable predictions about how often it occurs.

So when the MSM declares that a particular weather event falls into the “500-year” or “1,000-year” category, it’s often based on incomplete data, assumptions, and models that are far from definitive.

In my previous article, I highlight the problem of relying on limited datasets when making sweeping claims about extreme weather events.

I point out, that the high side tail takes the longest to fill in, because the extremes are, by definition, rare. This means that the further out we go on the distribution, the more speculative the claims become about the frequency of such events​

The High-Side Tail Takes the Longest to Fill In

Here’s an analogy that might help: imagine you’re filling a jar with marbles, but some marbles are much rarer than others.

Let’s say most of the marbles are white, but there are a few rare blue ones mixed in. You’ve been scooping marbles into the jar for years, and so far, you’ve only found a few blue marbles.

Someone might be tempted to declare that finding a blue marble is incredibly rare, maybe a “once in 1000 scoops” event. But if you’ve only scooped 50 times, that conclusion is, at best, premature.

The same principle applies to weather extremes. We’re dealing with a relatively short period of data collection, and because of this, we’ve barely begun to fill in the rare “blue marbles” of extreme weather events.

Yet, the media and even some scientists act as if we’ve already mapped out the entire distribution.

Insufficient Historical Data

The fundamental problem is that we simply don’t have enough real-time data over a long enough period to make robust claims about the frequency of such rare events.

Meteorological records only go back about 150 years in most regions, and high-quality, granular data is even more recent.

This is a far cry from the 500 or 1,000 years needed to reliably estimate the occurrence of these so-called extreme events.

Imagine trying to estimate the frequency of a rare weather event from a dataset that covers less than 1/3rd of the time required to make a “500-year” claim.

The statistical uncertainty becomes massive, and any declaration about a “1,000-year event” becomes almost meaningless in this context.

You’d need far more observations of such rare events to make even a modest claim with confidence.

As I detailed in a piece titled ‘Smoothing the past…‘, historical weather and climate data are often manipulated to fit a specific narrative. We see the same tendency in the reporting of extreme weather events.

The narrative tends to oversimplify, smoothing over uncertainty to deliver a dramatic headline without fully understanding the statistical limitations.

The Role of Models: Garbage In, Garbage Out

Now, let’s talk about the models that are often used to make these predictions. Models can be powerful tools, but they are only as good as the data and assumptions they’re built on.

When you start plugging incomplete or manipulated data into these models, the results can be wildly inaccurate. It’s the classic case of “garbage in, garbage out.”

These models often assume a level of certainty that simply doesn’t exist when it comes to rare weather events. The reality is that the more extreme the event, the less confident we can be in our predictions. But you’d never know that from the headlines.

How Many Events Do We Need?

To have any real confidence in claiming that an event is a “once in 500 years” or “once in 1,000 years,” you’d need a huge number of observations.

In fact, to even start talking about something being a 1000-year event, you’d want at least 1000 years of data, and even then, you’d want to have seen several instances of that event to build a reliable probability distribution. We’re nowhere near that level of data collection.

Media Sensationalism and Public Perception

So why does this happen? Why does the media latch onto these dramatic, but statistically unsound, claims? In part, it’s because fear sells.

A headline that says, “Heavy Rain in Wilmington” doesn’t get the same attention as, “Wilmington Hit by a Once in 1000-Year Flood.” The more extreme the event sounds, the more people will click, share, and discuss it.

But beyond just grabbing attention, this kind of sensationalism distorts public perception of what’s really happening with the climate.

If every storm, flood, or heatwave is framed as a “once in a lifetime” event, it’s easy for people to believe that the climate is spiraling out of control, when in fact, these events may not be as rare or unprecedented as they’re being made out to be.

In conclusion, the statistical foundations for claiming we’re seeing “500-year” or “1,000-year” weather events are weak at best.

We simply don’t have enough data to make these claims with any real certainty, and the use of incomplete or manipulated models only makes the situation worse.

Instead of jumping to extreme conclusions, we need to take a step back and recognize that the world is more complex, and far less predictable than these dramatic headlines would have us believe.

See more here Substack

Please Donate Below To Support Our Ongoing Work To Defend The Scientific Method

PRINCIPIA SCIENTIFIC INTERNATI ONAL, legally registered in the UK as a company incorporated for charitable purposes. Head Office: 27 Old Gloucester Street, London WC1N 3AX. 

Trackback from your site.