Cloud Seeding

by Brian Dunning
January 14, 2025

Humans have been trying to make it rain for millennia. We seem to always need more water, mostly to irrigate our crops, whether we are 21st century people building a highly technological dam or people 5,000 years ago doing a rain dance. Some 75 years ago, a researcher at General Electric discovered that he could use dry ice to seed supercooled water droplets and cause them to precipitate out of the atmosphere as rain. And ever since then, people have debated whether this is science or just another rain dance.

Certainly many have taken it for undisputed science. In preparation for the 2008 Olympics in China, the Beijing Weather Modification Office engaged in large-scale cloud seeding efforts, hoping to divert any possible rain during the Olympic Games opening ceremonies to fall elsewhere. Beginning four hours before the opening ceremony, from 21 locations they fired 1,100 small rockets loaded with silver iodide crystals into the clouds. The opening ceremonies ended up being mostly dry, and many hailed the cloud seeding as a great success.

In early 2024, the United Arab Emirates, which has long practiced cloud seeding to maximize the rain their dry country receives, was deluged with intense rainfall that fell across the entire Persian Gulf region. About 50 people were killed in flooding throughout the Gulf states. Unlike the praise that followed China's effort, it was angry blame that was broadly leveled against the UAE for causing the catastrophic floods.

These examples — plus countless others — would seem to be evidence that cloud seeding is effective at producing rain when and where it's needed; perhaps even too much of it. Yet the practice does not enjoy universal acceptance among atmospheric scientists. Some say it can increase the amount of rain by 30%; some say perhaps 1% at most. Results are virtually impossible to measure since rainfall amounts are never precisely predictable, and whenever we try a specific cloud seeding experiment, it's not possible to have a control for that experiment — once you seed a cloud, you can't compare the results to what you would have gotten if you hadn't seeded it.

Cloud seeding's highest potential comes when atmospheric moisture is supercooled. Supercooled water droplets are those that remain liquid water even though they are at subfreezing temperatures. This happens where there is no nucleus or surface or anything around which the water molecules can crystallize. But if they are disturbed, or come into contact with anything, they crystallize instantly. Whole clouds can become supercooled when the temperature is below freezing. It requires that the water droplets be very pure, with no contaminants that would serve as a nucleus. Such conditions are very dangerous to aircraft, as supercooled droplets instantly form ice on the leading edges of the wings.

Thus, supercooled clouds are the most obvious target for cloud seeding. Dropping any kind of suitable nuclei into them will form ice crystals, which catch other droplets and assimilate them as well. These then precipitate to the ground, eventually landing as snow or rain depending on the temperatures closer to the ground.

Today we typically use silver iodide crystals, or sometimes potassium iodide. Spraying these into the atmosphere poses no meaningful environmental risk, as it's not harmful to people or wildlife, and the quantities used are relatively minute and undetectable after the rain falls — far below natural levels. Also used is dry ice or even plain salt, ground very fine. When any of these particles contact a supercooled water droplet, you instantly get a tiny raindrop which will collect any other droplets it touches.

The basic problem with cloud seeding is that no matter what you try, you cannot create water where none exists. Seeding in a dry environment is doomed to failure because there is no water. But seeding into a sky where moisture does exist is achievable; however, any water you take out today is gone tomorrow. The best cloud seeding can hope to accomplish, even theoretically, is to borrow from tomorrow's rainfall. Bring it down today, and you'll have an even drier tomorrow.

With this in mind, experimentation sometimes focuses on places where "a drier tomorrow" makes no difference; for example, coastal areas where prevailing winds carry moisture-laden clouds out to sea. Upwind seeding in places like this aims to grab that moisture out of the clouds before it all goes out to sea.

Another related practice is called orographic cloud seeding — orographic referring to effects caused by the presence of mountains. Cloud seeding experiments near mountain ranges try to pull the moisture out of passing clouds where it can fall as snowpack on the mountains, thus ensuring an adequate water supply of meltwater to carry dependent communities through the summer.

Almost by definition, this would seem to happen at the expense of communities downwind from those same mountains, and it hardly needs to be stated that efforts to steal water from one region to benefit another is going to be controversial at best. Thus, cloud seeding programs nearly always face bureaucratic challenges as well.

Cloud seeding also faces opposition from people who consider it a form of geoengineering, to which they object mainly on ideological bases. However the distinction should be made clear. Cloud seeding is a short-term, highly localized activity, often targeting a single cloud on a single day. Geoengineering, on the other hand, refers to long-term efforts to affect the climate — two very different types of campaign. Cloud seeding is not a form of geoengineering.

The main challenge that cloud seeding faces, however, is neither bureaucratic nor ideological: it's the science itself. A 1999 paper in the Bulletin of the American Meteorological Society summarized the results of many published studies, and found:

During the last 10 years there has been a thorough scrutiny of past experiments using glaciogenic seeding. Although there still exist indications that seeding can increase precipitation, a number of recent studies have questioned many of the positive results, weakening the scientific credibility. As a result, considerable skepticism exists as to whether these methods provide a cost-effective means for increasing precipitation for water resources.

More recently, a 2010 paper from Tel Aviv University and published in the journal Atmospheric Research concluded:

Re-analysis of the cloud seeding experiment and operations in Israel shows that seeding has not produced the expected enhancement in rainfall… This suggests that seeding had little or no effect on total precipitation on the ground. These results are in agreement with those presented by Rangno and Hobbs (1995), and the [daily rainfall] calculations of Kessler et al., 2006, Sharon et al., 2008.

Those three studies mentioned are cited a lot in this field, and they all come to basically the same conclusions: that any effect cloud seeding might have is probably small, virtually impossible to distinguish from natural factors affecting daily rainfall, and lacking in strong experimental evidence.

But if it probably isn't very effective, what about those two popular examples at the top of the show? Why did China spend so much money cloud seeding at the Olympics, and why does the UAE (as well as a lot of other countries around the world) spend so many millions doing it every year? Let's look at each case.

It had indeed been very rainy in the days leading up to the 2008 Olympics opening ceremonies, and China had long been practicing cloud seeding. Even though this has been universally reported in the world press as a great success, that may reflect a propaganda success more than a scientific achievement. China very much had an interest in showing off their technological prowess to the world.

Their plan was likely to succeed. Meteorologists predicted only a 47% chance of rain on the day of the ceremony, and only a 6% chance of the feared downpour during the ceremony itself. So, it was probably not going to rain anyway. And it didn't.

For the 2024 flooding in UAE and elsewhere, it's true that they had done cloud seeding just before the rainstorm. But they have also done it for decades, almost always without any success at all. As far as this particular storm goes, the cause was a complex of thunderstorms called a mesoscale convective system driven by a massive low-pressure zone in the upper atmosphere combined with low pressure at the surface. This storm was going to nail the area whether anyone did any local cloud seeding or not. Its cause was major, spread over a huge area. The seeding is unlikely to have had any impact on it at all.

So why do so many continue cloud seeding, if it doesn't help? The answer is simple. People think anecdotally. If it occasionally rains after seeding, our native confirmation bias causes us to remember the times it worked, and forget or blur together the vast majority of the times it didn't. Often we see governments make decisions based on popular opinion and bureaucratic debate; rarely do we see them make decisions based solely upon untainted good science.

And that's really the best summary of the whole science of cloud seeding. It's plausible and it "works" (sort of) — just not enough to make any difference or to justify the expense. But confirmation bias, combined with laypeople's wishes and expectations, will generally eclipse the science in the decision making process of non-experts. In so many cases across all the popular sciences, you'll see this same upside-down process repeated; and when you do, you should always be skeptical.

This episode is dedicated to the memory of Eduard Peter (2011-2023), a fine young fellow who put his unbounded curiosity to good work by learning voraciously every single day. When Eddie left, he did so loved by all who knew him, each of whom was enriched by the experience; thus setting an example for us all to follow. Take a moment to think of Eddie — and how you too can be more curious, to learn more, and to love more.

By Brian Dunning
Follow @BrianDunning