Is artificial intelligence making weather forecasts better?

By Stephen Beech

Weather forecasts looking two to six weeks ahead are getting better — thanks to AI.

A new machine-learning approach has broken records for weather forecasting for the key time frame in emergency planning.

American researchers developed the model that learns to correct errors in existing forecasting models to make their output more accurate over the two-to-six-week time period.

When applied to a gold-standard forecasting model, the new method doubled forecasting accuracy over that time frame, placing first in a prominent international forecasting competition.

Scientists say the new method could provide farmers, energy operators and emergency services around the world more time to better prepare for storms.

Weather forecasts beyond two weeks are notoriously unreliable.

But the new technique, described in the journal Nature, improves weather forecasting beyond two weeks, beating climatologic averages between two and six weeks for the first time.

Study co-author Judah Cohen said: “Not only does this do better on time averaged forecasts for this range, which is already very useful for farmers and members of the energy industry trying to match supply with demand, but it also improves predictions of extreme weather events, whether it’s flooding or cold spells or something else.”

Cohen, a climatologist at Massachusetts Institute of Technology (MIT), added: “It’s really important to predict both, but extreme events might be the most impactful part of this for society.”

MIT professor Edward Lorenz first came up with the idea of chaos theory to describe how tiny changes in initial weather conditions lead to completely different outcomes.

Cohen said: “He came up with this theoretical two-week limit where beyond that, we just can’t predict weather because the noise in the system grows so much that it overwhelms any signal you start with.

“That had a strong influence on the field that may have held people back from exploring forecasting beyond the two-week range.

“I never really believed that completely. In theory the atmosphere can create an infinite number of patterns, but it likes to fall into these favorite patterns over and over again.

“Once one thing happens, I think you can predict the domino that will fall after that, and the domino that will fall after that.”

He says the best weather forecasts today come from either dynamical models, which numerically simulate how the atmosphere and oceans evolve over time, or AI models, which learn to make predictions from historical weather data.

AI models have shown improvements in predicting weather at short ranges, but are even worse than dynamical models at longer time ranges.

Cohen focuses on forecasting weather two to six week in advance, also known as subseasonal forecasting.

He said: “In most cases forecasting in the sub-seasonal ranges is no different than going to a casino and rolling the dice.”

Cohen credits his co-author Lester Mackey, whom he has been collaborating with for more than a decade, with the initial idea of using machine learning to reduce errors in forecasting models.

He said: “We built this wrapper — Lester calls it postprocessing.

“We know AI models have a lot of errors, so we thought maybe we can quantify those errors and remove them.”

While other forecasting systems do more simplistic error correction, the research team’s postprocessing layer uses AI to learn from a forecasting system’s mistakes based on historic data and shifts its forecasts.

More from this section

Businesses are embracing AI in calls, and so are scammers. How can you trust a voice on the line?

Study: AI coding tools do not write accessible code

+3

Team searching for Noah’s Ark builds 3D model of Turkish mound

In that time range, forecasters use so-called “probabilistic forecasts” where the likelihood of a range of possibilities if given.

Cohen said: “As more forecasts and data become available, our model adaptively improves its correction.”

The team used MIT Engaging, a supercomputing cluster hosted by MIT’s Office of Research Computing and Data, but the resulting model does not require a supercomputer to run.

To put their model to the test, the researchers submitted three models last year that included their post-processing layer to the European Centre for Medium-Range Weather Forecasts (ECMWF).

It is considered the best forecasting center in the world, and holds forecasting competitions between models each season.

The submission that won first place combined ECMWF’s forecasting model with another, and then applied the postprocessing correction on top.

That approach ended up doubling the original model’s predictive power in the subseasonal range.

The model offered an improvement over relying on climatologic averages for the first time.

Cohen said: “Up until now, no one had been able to show a model predicts weather better than long term averages for these sub-seasonal forecasts.

“They’ve pretty much all been worse than that — you could literally flip a coin and do better.

“People have questioned if it’s even worth investing in models at all for this time range.

“We consistently beat climatology averages with our models.”

The model also showed a “statistically significant” improvement in the ability to predict extreme weather.

Cohen said: “That’s one of the more important findings, because it’s so difficult to accomplish.

“When the ECMWF developed their own, non-AI, error-correction to their model, it got worse at predicting extreme weather.

“I think that’s because traditional error, or bias, correction is always trying to get to the mean, whereas extreme weather is on the fringe of probabilities.”

The team has shown their technique could be used to improve land forecasts in most regions across the globe.

Cohen said: “What’s nice about our model is it’s an add-on wrapper.

“We used it with the ECMWF’s dynamical model because that was considered the best, but it can be used with any model.

“It even improves models that are already bias corrected using another AI algorithm.

“So it shows a robust improvement. And it’s open source and low cost.”

He added: “It could also be useful for energy and emergency management.

“I’ve talked to energy officials who say, ‘If I’m told two days before that an ice storm’s coming, I don’t have enough time, but if I had two weeks, that could be enough to make adequate preparations.’”