The still-unrealized promise of commercial Earth science data

This article originally appeared in the Jan. 29, 2018 issue of SpaceNews magazine.

For the last few years, proponents of commercial satellite weather programs, both in industry and in Congress, have talked up the promise of government data buys from such systems. For a fraction of the cost of developing its own satellite constellation, government agencies like the National Oceanic and Atmospheric Administration could instead buy data from commercial systems to augment its weather forecasting efforts.

At the urging of Congress, NOAA started its first Commercial Weather Data Pilot program in 2016, awarding contracts that September to two companies, GeoOptics and Spire, with a total value of a little more than $1 million. They were tasked with providing GPS radio occultation data, measuring the diffraction of GPS signals as they pass through the atmosphere. Such data, collected by the U.S.-Taiwanese COSMIC network of satellites since their launch in 2006, have proven useful in increasing the accuracy of global weather forecasting models.

The original pilot contracts expired at the end of April 2017, after which NOAA spent several months analyzing the data they collected. The results so far, though, have been underwhelming.

“We have gone through one contract already with the radio occultation community, and we found that the data aren’t accurate enough or comprehensive enough yet to meet our observing requirements,” said Stephen Volz, NOAA’s acting assistant secretary for environmental observation and prediction, at a Jan. 17 event in Washington on space cooperation between the United States and the European Union.

Only one of the companies, Spire, actually provided data to NOAA by the end of the contract. GeoOptics suffered launch delays, with its first satellites not launching until July 2017.

NOAA is completing a report on that initial round of the pilot program. Speaking at a town hall session at the Fall Meeting of the American Geophysical Union (AGU) in New Orleans Dec. 12, Karen St. Germain, director of the Office of Systems Architecture and Advanced Planning at NOAA’s Satellite and Information Service, said she couldn’t disclose the details of the report, but emphasized the agency needed data that meets a specific set of standards.

“NOAA really is a requirements-driven organization,” she said. “Our observations that feed our weather predictions and warning really must meet certain thresholds of performance.”

In a statement to SpaceNews, Spire Chief Executive Peter Platzer acknowledged that the data the company provided to NOAA in the pilot program “was not comprehensive enough for global impact.” However, he added that Spire has improved its GPS radio occultation data system by “an order of magnitude” since the pilot ended, and expects to improve it again by another order of magnitude in the next few months.

Platzer added that an independent assessment of Spire’s data by NASA’s Jet Propulsion Laboratory confirmed its accuracy met both the requirements of the existing COSMIC satellite system — which he called the “gold standard” in GPS radio occultation data — as well as those for the COSMIC-2 satellites scheduled for launch later this year. (Spire spokesman Nick Allain clarified the scope of the report March 19, saying that Spire’s data met requirements for having low enough noise in its measurements compared to COSMIC-1 and COSMIC-2, but a subsequent study in the coming months will assess Spire’s data more broadly, covering thousands of GPS-RO profiles.)

“NOAA has shown global leadership yet again in working with the private sector in testing its capabilities to see how a public-private partnership can further the global competitiveness in weather forecasting of the United States,” Platzer said. “Spire is looking forward to continuing this partnership and augmenting NOAA’s capabilities in delivering a ‘weather ready nation.’”

Quality of data is not the only issue for NOAA, though. Volz said at the January event that prices for commercial satellite weather data were both higher, and more volatile, than expected.

“It has not been competitive because they are still trying to understand what exactly they are building,” he said of commercial pricing. “You can generate a price-per-occultation number, and they vary by a factor of 10 depending on what the vendor might say they can do and what we contract for and what the government solution might have been from a research mission.”

Preparing for round two

Despite the outcome of the first round of the Commercial Weather Data Pilot program, NOAA is moving ahead. “They are improving and we expect them to continue to improve,” Volz said. “We will be consistently, on a regular basis, reaching out to the community, seeing if there is a commercial solution for a NOAA observation requirement.”

NOAA is already planning a second round of the program, St. Germain said at the AGU meeting, which will again focus on GPS radio occultation data. NOAA will release a formal request for proposals in the spring and make awards before the 2018 fiscal year ends Sept. 30.

“Round two will be more comprehensive than round one,” she said. The initial round simply focused on gathering data, without any of what she called “operational performance thresholds” like latency, data security and reliability that would be needed if commercial data were part of an operational weather forecasting system.

Those issues will be part of the second round of the program. “We are intending to mature our approach to the purchase of radio occultation data from commercial vendors,” she said. Companies will be required to securely provide the data to NOAA, and the agency will more closely assess how the commercial data performs in its numerical forecasting models. The second round will also seek six months of data, versus the three requested in the first round.

Another potentially controversial issue that the second round will address is data rights. NOAA has traditionally freely shared its weather data with agencies in other countries, in accordance with what’s known as Resolution 40 of the World Meteorological Organization. That’s raised concerns in the past among commercial weather companies, since other agencies would be less likely to buy the data directly from the companies if they can get it from NOAA for free.

“We’ll also be tackling data rights, and getting licensing terms,” she said. “In this pilot, we’ll be looking to share data with our partners in the operational community.”

NOAA is also planning a third round of the commercial weather data pilot, she said, although they will likely move on from GPS radio occultation data to other types of data, such as microwave sounding of the atmosphere. St. Germain said NOAA will soon issue a request for information, or RFI, regarding the types of data that could be included in future pilot programs.

Other agencies consider commercial data

NOAA is not the only government agency interested in purchasing commercial Earth science data. NASA issued an RFI Dec. 5 seeking details from companies that operate satellite constellations that collect imagery and other data and are interested in selling it to NASA. The deadline for responses was Dec. 22.

“The question that we’re asking in NASA is, what value do the data products that come from your small satellite constellations — either raw data or products that you decide to generate — have to the government to advance our research, science and applications interests?” said Michael Freilich, director of NASA’s Earth Science Division, at the AGU town hall session.

NASA plans to evaluate the responses not for a future competition but instead to decide which companies to enter into negotiations with for blanket purchase agreements for data. Those agreements should be in place by the end of March.

The data purchases are intended to give NASA, and scientists supported by the agency, an opportunity to evaluate the data from those commercial systems to see if they are useful enough to warrant additional purchases in the future.

“Once we decide the data is good, and we want to more, we’ll figure out the best contractual mechanism to acquire that data,” said Sandra Cauffman, deputy director of NASA’s Earth Science Division. NASA may do additional RFIs in the future, she said, to get data on new capabilities from existing or emerging companies.

“It is not intended to be a one-and-done,” Freilich said of December’s RFI. “Depending on how this works out, we will modify, expand, and refine this. We all recognize this is a rapidly evolving field.”

NOAA’s European counterpart, though, is taking a wait-and-see approach when it comes to commercial weather data. Alain Ratier, director-general of Eumetsat, said at the Jan. 17 event in Washington that an unnamed American company approached him within the last couple of years about buying data to complement Eumetsat’s fleet of Metop polar-orbiting satellites.

“The price of the data for one year was above cost of the operations of the full Metop satellites,” he said. “Probably the offers will be different now; we will see, but at that time it was not competitive at all.”

Ratier said he will study how the NOAA and NASA data-purchase efforts work out, and expects Eumetsat’s 30 member states to weigh in on the benefits and costs of commercial satellite weather data. However, he emphasized the quality of commercial data was more important than the quantity of observations companies could provide.

“We need to understand what is available from the commercial sector in terms of quantity and quality, because it’s essential to understand that the progress of forecasts is not about more data, it’s about better data,” he said. “And at this point in time we have no evidence that the commercial sector can deliver better data.”