A few years ago, the National Oceanic and Atmospheric Administration (NOAA) began an ambitious effort to develop one of the world’s most powerful weather models — a system intended to deliver far more accurate forecasts and better protect boaters, coastal communities and anyone who relies on weather-sensitive operations. Now, after several major upgrades and new collaborations, NOAA is preparing to partner with outside experts to accelerate that vision.
NOAA’s recent steps include formal cloud partnerships with Amazon Web Services, Google Cloud and Microsoft, dramatically expanding public access to the agency’s massive environmental datasets. The agency also tripled its supercomputing capacity, bringing its infrastructure closer to the capabilities used by leading international forecast centers. In parallel, NOAA has begun sharing National Weather Service modeling code with the broader scientific community so researchers and private-sector developers can help improve the underlying models.
Most recently, in March, NOAA issued a request for proposals offering up to $45 million over five years to a technology partner capable of building the Earth Prediction Innovation Center (EPIC). EPIC is designed to be a collaborative “extramural” center that brings together government, academic and commercial talent to develop and deploy improved weather and climate prediction tools on a shared platform.
“You are basically unleashing the potential for researchers to advance the science, and for the private sector to unleash potential uses for the data,” says Chris Vaccaro, NOAA’s senior media relations specialist. “NOAA is a massive data agency. We collect satellite data, surface data — we literally collect billions of pieces of data every day. It’s just a tremendous amount of data, and you want it to be used in the most holistic sense possible.”
NOAA envisions EPIC functioning much like an App Store for weather and climate applications: a common framework of code and data available to many users, with room for innovators from a wide range of disciplines to design new products, test advanced models and develop operational tools that improve forecasts. The goal is to lower technical barriers so scientists, software developers and private companies can focus on innovation rather than building duplicate infrastructure.
“It’s hard to describe because it’s still in the formative stages,” Vaccaro says. “This is a way of not just limiting the expertise of weather modeling to NOAA, but also opening it up to community. We want all kinds of people with all kinds of expertise.”
One clear motivation behind EPIC is restoring and extending U.S. leadership in numerical weather prediction. For decades, some international models — notably those maintained in Europe and the United Kingdom — have delivered consistently strong results. In some cases, those centers ran fewer models on more powerful computers, while U.S. operations balanced a larger mission across smaller computing resources. By increasing computing power and enabling community contributions through EPIC, NOAA aims to run higher-resolution models that better exploit the agency’s vast observational data.
Improving model accuracy directly benefits maritime and coastal stakeholders. More reliable forecasts help marina managers, commercial fishermen, recreational boaters and emergency planners better anticipate storms, adjust operations and make safer decisions during hurricane season and other hazardous weather events. Enhanced models also improve inland forecasts, flood warnings and long-range climate projections that underpin economic and public-safety planning.
Many of NOAA’s EPIC-related actions trace back to the Weather Research and Forecasting Innovation Act of 2017, which directed the agency to prioritize improvements in weather data, modeling, forecasting and warnings to protect life, property and the economy. Subsequent congressional guidance in 2018 further encouraged NOAA to accelerate community-driven scientific and technological advances in prediction. Collectively, those mandates helped shape the agency’s move to open data, shared computing and collaborative model development.
Prospective partners had until May 11 to submit proposals responding to NOAA’s request. The agency expects to select a technology partner and announce the award by this fall. If the plan proceeds on schedule, NOAA anticipates EPIC’s extramural center will be well underway heading into 2021, with the long-term aim of producing what the agency describes as “the world’s most accurate and reliable operational weather forecast model.”
The EPIC initiative represents a shift in how weather science is advanced: from a mostly internal effort to a distributed, community-driven model that leverages public data, commercial cloud platforms and university research to iterate faster and at scale. For users who depend on daily and seasonal forecasts — from coastal operators to national emergency managers — that collaborative approach promises more precise warnings, better preparation and ultimately, a safer response to severe weather.
This article originally appeared in the June 2020 issue.