NEPTUNE model set to replace the US Navy's global weather forecasting system
The US Naval Research Laboratory (NRL) has developed NEPTUNE, an atmospheric prediction system that is set to take over from the Navy's current global weather model. According to the laboratory, extended testing showed better forecasts of low-level temperature, moisture and wind, especially over open ocean, along with improved tropical rainfall and cloud prediction.
One framework instead of several models
The Navy currently runs different models for different jobs. The Navy Global Environmental Model (NAVGEM) covers the whole planet, the Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS) handles higher-resolution regional forecasts, and a variant called COAMPS-TC is used for tropical cyclones.
NEPTUNE is intended to bring these scales under a single modelling framework. Its design is meant to make better use of modern high-performance computing, so that finer-resolution predictions can be produced more efficiently. The first operational version will replace the global system only. Regional and tactical-scale capabilities will be transferred separately, as part of a longer move towards a unified approach.
What the testing showed
The laboratory reports gains in near-surface temperature, humidity and wind forecasts over the open ocean, a region where observations are sparse and forecast errors matter for ship operations. Tropical precipitation and cloud forecasts also improved. NEPTUNE can additionally run at higher horizontal resolution than the existing global model, which opens the way to representing smaller features such as thunderstorms.
The announcement does not give verification statistics, such as error reductions or the baseline periods and metrics used. Nor has the work been presented through a peer-reviewed paper in the source. Researchers wanting to judge the size of the improvements will need those details from the project team or from later publications.
From research project to operations
Alex Reinecke, an NRL meteorologist and the project lead, began work on the system around 2015. He later led a team of about a dozen federal scientists and a similar number of contractors. The project moved through basic and applied research stages, then received extra support from the Navy's Oceanography and Navigation branch and the Office of Naval Research to speed up development and transition.
NRL handed NEPTUNE to the Fleet Numerical Meteorology and Oceanography Center in Monterey, California, in January 2025. That centre supplies meteorological and oceanographic products to Navy forces worldwide. Once the system is operational, its global products will reach forecasters and the fleet in phases.
A deeper atmosphere next
Most atmospheric prediction systems reach about 70 to 100 kilometres above the surface, depending on configuration. The NRL team is developing a deep-atmosphere version of NEPTUNE that would extend to 500 kilometres. That height would allow the neutral atmosphere to be coupled to the Navy's ionospheric model.
Reinecke describes the planned combination as the first coupled neutral-atmosphere and ionosphere prediction system of its kind. That claim comes from the project lead and has not been independently assessed in the source. Even so, the ionosphere influences radio propagation and satellite signals, so linking it to the lower atmosphere is a long-standing goal in space-weather and atmospheric research.
Why it matters
For the wider modelling community, NEPTUNE is an example of a nonhydrostatic, multi-scale design built to scale with computing power, an approach several national forecasting centres are also pursuing. Open questions include how the regional and tactical components will be transferred, how the deep-atmosphere version performs once coupled to ionospheric prediction, and how the full verification results compare with NAVGEM. Those results, along with the planned phased roll-out, will show how far the reported gains hold in routine operational use.
Source: NEPTUNE poised to replace Navy's global weather model, improving forecasts over open ocean (Phys.org, 20 November 2026).
Image: AI-generated illustration.
Working on this topic? Find a journal for your research or join the discussion on Scholars Post.


