Temporal and spatial scales of extremes of daily tropical rainfall : the example of Thailand (1979-2024)

The Weather's Wild Card: Why Rain is Hard to Pin Down

Imagine trying to catch a specific raindrop in a storm with a net. That's roughly what scientists do when they study extreme rainfall. Rain isn't just a steady drizzle; it's a chaotic mix of tiny, short-lived thunderstorms (like individual sparks) that sometimes group together into massive, organized systems (like a roaring bonfire). This field of study, known as meteorology, tries to understand how these "sparks" and "bonfires" behave across space and time.

To make sense of this chaos, researchers use a few key tools. First, they look at percentiles, which are like ranking a race. Instead of asking "How much rain fell?", they ask, "Was this day wetter than 99% of all other days?" This helps them spot the true extremes without getting confused by how much rain usually falls in a specific place. Second, they measure spatial scales, which is just a fancy way of asking: "If it pours here, how far away does it also pour?" Is the rain a tiny, isolated puddle, or a giant flood covering a whole city? Finally, they check temporal scales: "Does this storm last for an hour, or does it keep raining for days?" Understanding these patterns is crucial because extreme rain can cause flash floods, destroy crops, and disrupt lives. If we can predict how big and how long these storms last, we can build better cities and prepare for the worst.

---

The Great Rain Hunt in Thailand

In this study, a team of scientists decided to play detective with rain in Thailand. They wanted to figure out exactly how big and how long the wildest rainstorms get. They didn't just guess; they looked at a massive dataset covering 46 years, from 1979 to 2024. To get the most accurate picture, they used two different types of "rain detectors." The first was a network of 75 real rain gauges scattered across the country—physical buckets that catch the actual water. The second was a set of three high-tech computer models (called MSWEP, CHIRPS, and ERA5) that estimate rainfall over the whole country using satellites and weather patterns.

The researchers were looking for the "ExtD" (Extreme Daily Rainfall). They defined these as days that were wetter than 90%, 95%, 99%, or even 99.9% of all other days. Think of it as finding the top 10%, top 5%, top 1%, and the super-rare top 0.1% of rainy days. They asked: How far apart do these extreme days happen? Do they last just one day, or do they stick around for a week? And, most importantly, can the computer models see these extremes as clearly as the real rain gauges?

The Findings: The Rain is Smaller, Shorter, and Getting Wilder

Here is what the team discovered, and it turns out the computer models have a few blind spots.

1. The "Pixel" Problem with Computer Models
The scientists found that the computer models are generally good at guessing the general weather, but they struggle with the really wild stuff. When a real rain gauge recorded a massive downpour, the computer models often said, "Oh, that was just a heavy shower." The models tended to underestimate how much rain fell on the most extreme days.

Why? Imagine a photo taken with a low-resolution camera. If you zoom in on a tiny, bright spark, the camera blurs it out and makes it look like a dull, larger glow. Similarly, because computer models divide the world into a grid (like pixels on a screen), they smooth out the tiny, intense bursts of rain. The real rain gauges showed that extreme rain is often very local—sometimes happening in one spot but not even 10 kilometers away. The models, however, made the rain look like it covered a much larger area than it actually did.

2. How Big is the Storm?
The team measured the "size" of these storms using something called an e-folding distance. This is a fancy way of asking: "How far do you have to travel before the chance of seeing the same extreme rain drops to about 37%?"

• For the "moderate" extremes (the top 10% of rainy days, or p90), the real rain gauges showed these storms usually cover a circle about 77 kilometers wide.
• For the super-rare, "monster" storms (the top 0.1% of days, or p99.9), the real storms are tiny! The study couldn't even measure a specific size for these because they are so small and isolated that the rain gauges were too far apart to catch them both.
• The computer models, however, kept saying these storms were huge. For the monster storms, the models guessed they covered areas up to 274 kilometers wide. The models were essentially seeing a giant cloud where there was actually just a tiny, intense spark.

3. The "One-and-Done" Nature of Rain
The researchers also checked how long these extreme days last. They found that most extreme rain days are "time-isolated."

• About 55% of the "moderate" extreme days happen alone, with no similar rain the day before or after.
• For the "monster" storms (p99.9), 90% of them are isolated single days.
• Even when these extreme days are part of a wet spell, the spell doesn't get much longer. A storm that hits the top 0.1% threshold doesn't necessarily mean it will rain for a week straight. The computer models tended to overestimate this, suggesting storms lasted longer than they actually did.

4. The Rain is Getting Intense
Finally, the team looked at the long-term trend from 1979 to 2024. They found a clear signal: Rainfall is intensifying.

• The total amount of rain in Thailand hasn't changed drastically, and the number of rainy days hasn't gone up.
• However, the frequency of the extreme days is shooting up. The number of days hitting the top 95% and 99% thresholds is increasing faster than the average rainfall.
• This suggests that while it might not rain more often, when it does rain, it's getting much heavier.
• Interestingly, the computer models had a hard time catching this trend. The ERA5 model did a decent job of showing this intensification, but the MSWEP model failed to capture it accurately. In fact, for the rarest storms (p99 and p99.9), MSWEP showed a negative trend, suggesting a decrease in extreme rain, which contradicts the observed reality. This tells us that if we want to study how climate change is making storms worse, we can't just rely on any computer model; we have to be very careful about which one we use.

The Bottom Line

This study is a reality check for how we track extreme weather. It shows that real-world rain is often more chaotic, local, and short-lived than our best computer models suggest. The models tend to blur the edges, making storms look bigger and longer than they really are. While the models are great for general weather, they struggle to capture the true "monster" storms that cause the most damage.

The most exciting takeaway is the trend: Thailand is seeing more of these intense, isolated downpours. The rain isn't just falling harder; the extremes are becoming the new normal. As the scientists noted, this suggests a long-term intensification of rainfall, a pattern that fits with what we expect from a warming planet. But to see this clearly, we need to trust the ground truth—the actual rain gauges—more than the smoothed-out pictures from the computer screens.