Karen St. Germain’s office in Silver Spring, Maryland, is a weather geek’s dream, with large windows providing expansive views to the west, across the northern suburbs of Washington.

It’s a fitting angle — ideal for watching incoming storms — for someone who occupies one of the top positions in the National Oceanic and Atmospheric Administration at a time the agency is caught between growing demand for timely and accurate weather information, and competition from a host of new companies threatening to beat government agencies at their own game.

The outcome of that competition could affect the public’s access to the best available weather and climate data in the years ahead.

Fueled in part by climate change, extreme weather is an increasing liability to the economy, with 10 weather and climate disasters costing more than $1 billion each so far this year, according to NOAA. During the past two years alone, Western wildfires have cost more than $40 billion. Hurricanes are dumping more rainfall than they used to, and heat waves are more intense and frequent.

Those rising costs — along with advances in data-gathering and processing, and cheaper access to low Earth orbit — have spurred start-ups and established companies to get into the business of weather forecasting.

Private weather forecasting is a $7 billion industry and growing, according to a 2017 National Weather Service study. It’s also increasingly testing the federal government’s hold on weather data and warnings.

Those pressures are expected to grow as forecasting moves into environmental prediction, such as anticipating harmful algal blooms and dengue virus outbreaks. The Trump administration has so far shown little inclination to make sure government agencies stay ahead of private competition.

Until recently, AccuWeather, Earth Networks, the Weather Co. and other private weather providers relied on the fire hose of data from NOAA’s National Weather Service and satellite arm, as well as NASA and other agencies. Now companies are producing their own data and using analytics in business-savvy ways, tailoring their forecasts to specific real-world problems.

With the ability to launch satellites and supercomputers and to harvest data from semiautonomous vehicles and wearables, the new arrivals are leapfrogging the information-gathering capabilities of federal agencies.

They are also more nimble in analytics, using machine learning, artificial intelligence and cloud-based systems to warn a railroad company when to avoid a tornado barreling toward a specific stretch of track, or a farmer when to irrigate a particular row of crops. These companies are telling airline ground controllers when they might need to de-ice planes, or reschedule flights to avoid severe thunderstorms.

And they are putting the National Weather Service in an awkward position as it tries to fulfill its mission of protecting lives and property. The agency faces the prospect of having to partner with outside companies to get the best data. Not all of them are willing to share. Some of them harbor ambitions of taking over more of the federal government’s functions.

Growing tensions

As the director of the Office of Systems Architecture and Advanced Planning at NOAA’s Satellite and Information Service, St. Germain, a satellite instrument specialist and graduate of the Naval War College, is charged with navigating through that unfamiliar business environment, and ensuring the country has the data it needs to prepare for the extreme weather and environmental events headed our way.

Among her challenges are the growing tensions between, as she sees it, two ends of the value chain when it comes to weather and climate observations.

“So on one hand we’ve got the emerging interests that want to make observations and sell them,” St. Germain said. “And they want to, of course legitimately, then they want to control the licensing terms and so forth and sell to more than one [customer]. And then on the other end of the value chain, we’ve got the folks who make their money by building tailored products.”

Those companies are used to getting their data free from the government and using it to create their products. “So there’s a lot of natural tension there. And I don’t know how all of that will play out,” she said.

In such a fast-moving business environment, there are clear risks in spending public money on novel technologies.

St. Germain, who is powered by an espresso machine on her desk adorned with Washington Nationals bobble heads, has to make sure the data is accurate and reliable — and that any company she partners with will remain in business for the long term. She is worried that NOAA would contract with a private firm for data that is crucial for issuing timely severe weather warnings, only to see that company suddenly go out of business.

It is, she said, a balancing act.

“We don’t have to frame this as an either/or question. We don’t have to choose between relying only on commercial data sources or only on traditional ‘government’ sources,” she said. In President Donald Trump’s fiscal year 2020 budget request, she said, there is an increase to buy space-based commercial weather data, which would be a first for NOAA.

“It could be at some point in the future that commercially provided observations are as prevalent as phone service or any other service you buy,” she said.

But that point has not yet arrived. “We don’t yet know how many commercial suppliers the market will sustain, nor do we have an understood price structure. It is a real challenge to factor that uncertainty into NOAA’s mission continuity needs,” she said.

Some of the companies vying for NOAA data contracts may not be turning a profit, which raises questions about their viability. That would need to be factored into the government’s calculations.

For now, NOAA is the only authorized issuer of severe weather watches and warnings in the country, and it still is widely viewed as the leader in accurate weather forecasts and lifesaving warnings.

St. Germain said that responsibility can’t change — no matter how many new entrants get into the weather business. “The fundamental public good predictions and warnings, I think is one that has to stay with NOAA, because it can’t be subject to the profit motive,” she said.

But the expansion of private weather forecasting means we are careening toward a future in which the federal government may no longer be the primary source of weather information.

Consumers are already getting used to the idea of private sources of information for weather — even if the data originates with the Weather Service. We rely on weather apps to get our daily forecast, be it Weather Underground, the Weather Channel, Dark Sky or many others. It might not be that big of a leap for people to get used to paying for a subscription service to get a basic forecast or even severe weather warnings, much like we pay for our favorite shows on Hulu or Netflix.

Flurry of launches

The past decade has seen a flurry of launches as well as consolidation in the private weather business, with IBM’s $2 billion purchase of the Weather Co. in 2016, which included the popular weather.com website and app but not the cable network attached to that site. Climate Corp., whose computer modeling tools provide farmers with the ability to plan for increasingly common weather whiplash, with the Midwest lurching from drought to flood and back again, was bought by Monsanto in 2013.

In November, IBM rolled out a global weather forecast model developed in partnership with the nonprofit National Center for Atmospheric Research that claims to accurately predict small-scale weather features such as severe thunderstorms. IBM says the service represents an advance for places like Africa and South Asia that may have few weather observation posts and poor infrastructure. For now it’s only a short-term forecasting tool, covering out to 12 hours. But the model could yield huge gains for government warning systems in nations that have struggled to warn their people in time of approaching severe weather, including tropical cyclones.

Spire, a start-up, has launched a fleet of dozens of tiny satellites that track ships and aircraft, and gathers atmospheric data to help improve weather predictions. The company is marketing that data, known as radio occultation, to NOAA and international agencies as a way to improve their forecasts or eliminate the need for their own larger, more expensive satellites.

Ursa Space Systems, which uses satellite-based systems to provide images of Earth’s surface even when there is cloud cover, is producing products aimed at the insurance industry. Descartes Labs and Planet Labs are using imaging analysis for fire weather prediction. They, along with other satellite imaging firms, are also competing to sell data to financial firms to predict company earnings.

ClimaCell, founded by three Israeli military veterans who went on to graduate from Boston-area business and management programs, uses data harvested from cellphone towers, vehicles and other unconventional sources. JetBlue has invested in the company and uses its tools for forecasting operations at major hubs including Boston and Fort Lauderdale.

“There’s a ton more value to be created that does not try to cannibalize or does not compete” with government, said Rei Goffer, co-founder and chief strategy officer. ClimaCell is using cloud computing to branch out from “micro-weather” forecasting, such as making a forecast for the next hour at an individual airport, to developing a computer model that can go out to two weeks. The company plans to extend that time frame beyond that, potentially into seasonal forecasting, Goffer said.

He does not necessarily see that as competing with the government. “It’s about how much more on top of what governments are already doing, can you do, and how much value are you creating?” he said.

Climate change

These challenges to the primacy of the National Weather Service come at a time when Americans increasingly are focused on the weather because of climate change and the growing impact of extreme weather.

“Weather has moved into this much bigger role in people’s psyche today because of the amount of devastation that it’s bringing,” said Washington Sen. Maria Cantwell, the ranking Democrat on the Senate committee that oversees NOAA and NASA. Cantwell said she wants NOAA to have the technology it needs to better serve the public rather than ceding ground to the private sector. “We shouldn’t be playing second fiddle to anybody on weather forecast information,” she said.

The Weather Service is already sometimes lagging behind global technological advances, especially in computer model accuracy.

Europe’s main weather forecasting model often beats the Weather Service’s Global Forecast System — most famously when the European model accurately predicted Hurricane Sandy’s unprecedented track a week in advance in 2012.

Europe’s plans for increasing its computing power for the next decade would keep it the leader, in part because the agency that runs the model charges for the data, rather than giving it away as NOAA does.

The National Weather Service, with congressional support, is trying to close the gap, chiefly through a new modeling program called the Earth Prediction Innovation Center, which aims to speed up model development through collaboration with universities, other agencies and the corporate sector.

According to Mary Glackin, a veteran of senior-level positions at NOAA and IBM who is president-elect of the American Meteorological Society, the agency isn’t innovating quickly enough.

“When you look at a flash-flood warning, it looks about the same as it did 25 years ago,” Glackin said. “You kind of know a whole lot more about where your Lyft or Uber driver is and when he’s going to get to you than you know about any flash flood in relation to your geography.”

Andrew Blum, author of the book “The Weather Machine,” warns that commercial competition for data could put the public at risk if agencies like NOAA allow it to stifle the flow of free data, which is fed into every global computer model.

“The technology of the weather forecast has changed dramatically over the past decade,” Blum said.

“None of these models work unless we have the global data,” he said. The risk we’re running now is that “someone turns off the spigot,” possibly motivated by legal agreements stemming from commercial data buys, robbing everyone else of key data.

Public-private efforts

Some companies, like IBM, are cooperating with the public sector by putting some of the computer code they are developing back into an open-source system. Others are seeking to sell their data to the government, much as Northrop Grumman sells its drones to the Pentagon.

Such “data buys” would come with use restrictions — namely, that the public and possibly international partners would not be able to tap into it.

Scott Rayder, a former chief of staff to the NOAA director under President George W. Bush and currently the senior adviser to the president of the University Corporation for Atmospheric Research in Boulder, said the technological developments raise important questions about the government’s core function of issuing warnings and protecting its citizens from extreme weather, climate and environmental events.

“There’s so much data out there that can be determined to be weather data,” Rayder said. “And now companies are getting smart about how do you source that data and get value out of it.”

But that raises equity issues.

“I don’t want to get into a world, frankly, where if you have more resources, you can get a better forecast,” Rayder said. “There’s got to be a minimum level of warning and forecasts to protect life and property.”

Glackin said there’s broad agreement on a continued need for reliable observations for private companies and the government, and these would require public investment. It also makes sense to leave the warning function with the government, she said.

“The NWS is one of the most trusted parts of the federal government, so would [citizens] trust a warning coming from AccuWeather? I think not, I don’t think it’s the same type of thing,” she said. “Abdicating this to a private company and their interests makes no sense to me.”

So far, the Trump administration has been sending mixed signals about where it wants to steer the public-sector weather enterprise, as NOAA moves ahead with experiments to buy weather data from private companies.

In 2017, Trump nominated Barry Myers , the former CEO of AccuWeather, a company that has advocated for the privatization of large parts of the National Weather Service, to lead NOAA. However, his nomination stalled, in part because of conflict-of-interest concerns since his family still owns the private forecasting company. Myers asked the White House to withdraw his nomination this month, citing health concerns.

Trump has shown a willingness to elevate political interests above forecast accuracy and climate science. In September, he forced NOAA to issue a statement backing up an inaccurate presidential tweet that stated Alabama would be severely affected by Hurricane Dorian. That action, which criticized the Weather Service’s Birmingham, Alabama, office for issuing accurate information, delivered a credibility hit to the weather agency.

Then there’s climate change, a subject that puts Trump at odds with the global scientific community, let alone NOAA, as he continues to deny the reality of human-caused warming.

As the climate continues to change, government agencies like NOAA, as well as private-sector companies, will be tasked with an entirely new array of functions. That could significantly stress the forecasting system, if it’s done the wrong way. On that, the new weather entrepreneurs and government scientists agree.

“Governments are still going to be focused on protecting lives and property,” said ClimaCell’s Goffer. “A future where it costs you money to get a hurricane alert is a bad future, and we shouldn’t be aiming for that.”