Reading The Sky

Our ability to predict the weather has advanced gradually for decades, with great benefit to farmers. How far out can we get?

By Boyce Upholt | Photos By Mike Hollingshead

Since the dawn of agriculture, farmers have noted signs suggesting the next day’s weather, from the color of the clouds to the speed of the wind. Heat and rainfall arrived in seasonal patterns, which allowed crude projections about when to plant and harvest. Beyond such simple calculations, though, predicting the weather was often the domain of priests—and sometimes charlatans.

But at the turn of the 20th century, a few savvy thinkers realized the laws of physics, combined with the power of mathematics, offered a clearer picture. Over decades, the accuracy of “numerical weather prediction” has advanced slowly but steadily. In a recent paper in the scientific journal Nature, a group of meteorologists called it a “quiet revolution,” one that has, almost beyond notice, completely changed how farmers and others plan. Now, new legislation, signed by President Trump in 2017, is calling for improvements in forecasting, especially long-range forecasts, looking forward two weeks and beyond. 

Numerical weather prediction depended on three historical breakthroughs. First, over centuries, new tools were invented—thermometers, barometers and so on—that replaced visual observations with hard-and-fast data. Then, equations were developed that describe changes in the atmosphere. Finally, computers emerged, which were able to process those equations at high speed. By the middle of the 20th century, short-term weather forecasts began to feel routine. 

Still, advances have continued. New satellites and weather stations provide increasingly robust data, and computers grow more powerful. Generally, we gain one day of forecasting every 10 years. Today’s six-day outlook is about as precise as a five-day forecast was a decade ago. 

Looking Long Term

The Climate Prediction Center (CPC), a part of the National Weather Service, looks out even farther. It has been making three-month-long seasonal forecasts since 1995. These “climate” predictions—so called because weather, over the long run, becomes the climate—require different inputs than shorter-term forecasts.

“We’re looking for data that has a kind of memory, that holds over time,” explains Jon Gottschalck, the acting chief of the operations branch at the CPC. For instance, ocean surface temperatures are especially useful, as these change slowly and have significant impacts on the atmosphere. There are “oscillations” in regional ocean temperatures—which alternate between occasional warming and cooling periods—that recur with some regularity and help provide predictability over a seasonal scale. The most famous of these is the El Niño–Southern Oscillation, or ENSO. (Nonscientists often call this El Niño, the name for its warming period.)

Surprisingly, the “subseasonal” time period—from two weeks to three months—remains a murky zone. It’s far enough into the future that equations based on current atmospheric conditions lose accuracy, but a bit too quick for oceanic data to work its magic. The first consistent monthlong forecasts were not produced until the late 1990s in Canada. The CPC did not offer subseasonal forecasts until 2015.

During the past two decades, scientists have been chipping away at the problem. New satellites pick up more and more details, and faster computers can handle more complex climate models. 

Scientists also are learning to better model atmospheric processes that have impacts at the subseasonal scale. The Madden-Julien oscillation, another tropical pattern that repeats more frequently than ENSO, is the most valuable source of predictability for such forecasts in the United States. Meteorologists also consider the impact of other slow-moving data sources, such as snow cover and soil moisture, on the atmosphere.  

New Motivation

The Weather Research and Forecasting Innovation Act of 2017, H.R.353, signed into law in April 2017, was the first major piece of weather-forecasting legislation in decades. Republican Sen. John Thune of South Dakota, who chairs the Senate committee overseeing the National Weather Service, made seasonal and subseasonal forecasting—together called S2S—his top priority. 

This was in large part because of the potential impacts on farmers. “Effectively preparing for Mother Nature’s curveballs is critical for farmers and ranchers,” Thune told FarmLife. Among other provisions, the bill authorized $26.5 million to be spent on improving such forecasts in 2018 alone.

The CPC considers farmers one of its highest-priority “end users,” and already works closely with USDA, especially through the department’s 10 Climate Hubs around the country. Dennis Todey, director of the Midwest Climate Hub in Ames, Iowa, says one of the biggest issues is simply getting farmers to consider long-range outlooks. 

According to survey data he’s consulted, daily forecasts are used nearly universally by farmers—but the farther out the forecast, the less it is viewed. Projections beyond 30 days are almost never consulted. “People either don’t know how to use that information, or they don’t trust that information,” Todey says.

That’s in part because seasonal and subseasonal forecasts look quite different than the nightly TV weather report. Rather than a snapshot of future conditions, they offer a probability distribution. The forecasts might say, for example, that there’s an 80% chance it will rain more than usual during the next few months—while offering nothing about how much more. 

Still, such probabilities can be useful; knowing that a season will be particularly hot or particularly wet can help clarify what crops and/or varieties to plant. It also can inform farmers’ crop yield estimates and, therefore, financial decisions. Todey’s team at the Midwest Climate Hub aims to make this information even more useful to farmers by offering a two-page interpretive document, with text to accompany the visual forecasts, every few weeks. The document includes a drought monitor alongside the forecasts, which helps farmers consider how coming precipitation may alter current conditions.

Stretching Farther

Gottschalck at the CPC believes that during the next few years, subseasonal forecasting will offer increasing detail. Scientists may be able to predict, for example, the number of days that will be above 90°F in the coming month. Farmers can expect more precision when it comes to extreme weather, too, including earlier predictions about when a hurricane might make landfall or the occurrence of flash floods. Monthly forecasts will be extended out to 45 or even 60 days. 

Yet, snapshot precision will likely never be possible on a long-term scale. “We’re not going to wake up one day and go, ‘Oh, we can tell you what’s going to happen a year from now,’” Gottschalck says. “No, that’s not going to happen.”

For now, the biggest breakthroughs may not be meteorological. In 2016, the CPC partnered with the University of Maryland to conduct surveys and focus groups, evaluating how users, including farmers, utilize its predictions. As a result, the CPC is considering implementing changes, such as adjustments to the map legend design and explanatory text—somewhat similar to modifications being made by Todey’s team. The National Oceanic and Atmospheric Administration, which oversees the National Weather Service, has indicated to Congress that continued behavioral research is a top priority.

The scientists who spoke to FarmLife were enthused to see congressional support for improved forecasting. “[Now] there’s a firm commitment from our government,” Gottschalck says. “It’s a public law, and that sort of law does give us a mandate to try to work toward better products in that area.” 

Yet, it’s not as if a signature on a bill can suddenly force a breakthrough. The new bill is filled with what Gottschalck calls “embellished language.” 

“That’s fine, they can do that,” he says. “[But] all of our reports so far have made clear right up front that improvements have always been very incremental, and that’s likely to continue. It’s unlikely to have a silver bullet.”