Dubai turns to AI to make it rain: Here's what the revolutionary plan is for "seeding" clouds

In the marble halls of a luxury hotel, artificial intelligence experts sketch out modern solutions to an age-old problem: how to make it rain in the United Arab Emirates, a wealthy nation in one of the world's largest deserts.

This Gulf country has spent decades working and millions of dollars trying to mitigate its dry, hostile and terribly hot climate, which, however, has not prevented the growth of a predominantly expatriate population.

Results have been sparse so far. But at the International Precipitation Enhancement Forum held last month in Abu Dhabi, the idea of ​​using AI to extract more moisture from often clear skies was floated.

One such initiative is an AI system to improve cloud seeding, a technique that uses aircraft to spray salt and other chemicals into clouds to increase rainfall.

The system is "almost finished," said Luca Delle Monache, deputy director of the Western Center for Weather and Water Extremes (CW3E) at the University of California, San Diego.

"We are putting the finishing touches on it," he added.

The scientist admits that AI is not a "miracle solution" for the UAE.

Cloud seeding increases the size of droplets in clouds, which then fall as rain . It is estimated that it increases precipitation by 10-15%, he explains.

But it only works with certain types of clouds, and if not applied correctly, it can even prevent rain.

"It has to be done at the right time and in the right place. That's why we use artificial intelligence," he says.

The three-year project, funded with $1.5 million from the United Arab Emirates, feeds weather, satellite and radar data into an algorithm that predicts where seedable clouds will form in the next six hours.

Currently, a group of experts studying satellite images is in charge of directing the hundreds of cloud seeding flights that take off every year in the country.

With only about 100 millimetres of rain a year, the UAE's nearly 10 million people rely mainly on desalinated water, piped in from plants that produce 14% of the world's total, according to official data.

The population is 90% foreign and has multiplied almost 30-fold since the country was founded in 1971. The inhabitants are concentrated in large cities such as Dubai, Abu Dhabi or Sharjah, coastal refuges in the vast Arabian desert.

However, the country needs rain to feed the water table and a series of dams used for agriculture and industry.

Although authorities say the rainfall has increased, it is so unusual that when a few drops fall, children start clapping and run to their classroom windows to see this unusual phenomenon.

Rain, even artificial rain, has become an attraction: on Dubai's Raining Street, visitors pay 300 dirhams ($81) to walk in fake drizzle.

Last April was an exception. Record rainfall flooded Dubai, shutting down its airport and paralysing the city for days.

To find solutions, the UAE began organizing this forum of experts in 2017, and its rain improvement program has provided $22.5 million in scholarships over a decade.

"It's a very niche area in atmospheric science. There are only a few experts in the world and they're almost all here," says Delle Monache.

The algorithm his team is preparing is not the only use of AI discussed on the forum.

Marouna Temimi, an associate professor at Stevens Institute of Technology in New Jersey, presented a U.S. system that uses machine learning to track the path and impact of storms in real time.

But like Delle Monache, Temimi is also cautious about these AI-based solutions and their limits.

The lack of detailed data on cloud composition — a common problem because the equipment to analyse it is expensive — makes accurate predictions difficult even for AI, he says.

"We still have work ahead of us because we have data, but not enough to train the models correctly," he explains.

World Water Council President Loic Fauchon also called for caution and finding the right balance "between artificial intelligence and human intelligence."

"Let's not go too fast with artificial intelligence. Humans are probably the best option," he said.