The Next Generation of Marine Forecasting

Photo Credit: John Cobb

It’s no secret that present technologies have rapidly evolved as we try to push the limits of computing. While these advancements are often most apparent in the releases of the newest smartphone or self-driving car, our numerical weather prediction models have undergone exciting advancements as well.

These numerical weather prediction models once ran exclusively in large, central computing facilities. Now, however, they can be run locally, perhaps even on the device with which you are reading this story. This dramatic decentralization of forecasting computing has resulted in an explosive increase in the use of local high-resolution models throughout the past decade. Known as SWAN, these models have resolution capabilities beyond similar datasets, and can be used to visualize wave heights, currents, and more variables for marine and coastal industries.

The SWAN Model Now in the Planet OS Datahub

Furthermore, we are excited to announce the release of the new hyper-local wind & wave forecast datasets based on the SWAN model as a part of continuous Marine Data support in the Planet OS Datahub.

Coastal communities, businesses, and professionals significantly rely on the quality of the marine weather predictions. From major weather events such as typhoons and hurricanes to a quick check for the best time to go to the beach, reliable marine weather models affect the lives of millions. Thus, with this release, we hope app developers and domain experts will continue to create value-add applications and extend their customer base in marine navigation, coastal tourism, aquaculture, and other related domains. We also encourage businesses to utilize such data in their analytics to derive valuable insights for operations, planning and risk assessments.

Defining SWAN and NWPS

Nearshore Wave Prediction System (NWPS) provides on-demand, high-resolution nearshore wave model guidance to U.S. coastal Weather Forecast Offices (WFOs), triggered in real time by forecast wind grids prepared and submitted by the individual offices. NWPS is driven by forecaster-developed wind grids produced in AWIPS, and wave boundary conditions from the operational WAVEWATCH III model. The model core used is the spectral wind-wave model SWAN. Wave-current interaction is included using surface currents from the Real-Time Ocean Forecast System (RTOFS-Global).

NWPS is a downscaled, coastal wave modeling system for providing wave forecast guidance to coastal Weather Forecast Offices (WFOs) that are consistent with their marine wind forecasts. It is currently implemented in NWS Southern and Eastern Regions, and the National Hurricane Center (Tropical Analysis and Forecast Branch).

NWPS is also an on-demand modeling system, forced by forecast wind fields and submitted by WFOs via a baselined GUI in AWIPS. Additional forcings from other NCEP models include: wave boundary conditions from WAVEWATCH III, water levels from ESTOFS (extra-tropical), P-Surge (tropical), and surface currents from RTOFS Global. The system provides 102-hour forecast guidance for two cycles a day (on average). Now on Planet OS, we provide the SWAN model for 13 areas. However, some are not applicable but can easily be updated upon request. Currently, the first model, ATM (Key West) is in operation. All models are currently within the Southern Region of US as defined by NOAA so far and can also be added upon by request.

Basic forecast products include integral parameters of wave fields, partitioned and tracked wave systems, wave spectra, and statistical wave run-up and rip current guidance (currently at pilot sites only).

SWAN Model of Wave Height

Why SWAN Modeling?

While the SWAN model doesn’t cover areas as large as global models like WW3, its spatial resolution is nearly 17 times higher than the WW3 global model. This resolution is especially useful in forecasting as the scale of open ocean processes are much more localized, especially regarding shore-based activity. Where in a WW3 model one can see a lower resolution model of large scale, global phenomena, SWAN allows for higher resolution, clearer visualizations of location specific phenomena.

WW3 Model (Left) resolution compared to SWAN (Right)

In a plethora of cases where it is vital that our weather forecasts issued over the coastal waters be as accurate as possible, SWAN is a means to improve the guidance available when producing marine forecasts. With the ability to run a high resolution nearshore wave model on demand, it provides an immediate service enhancement to developers, businesses, and communities alike.

DIY NOAA Nearshore Wave Prediction System SWAN Wave Model

If you are curious about or would like to learn more about the NOAA Nearshore Wave Prediction System SWAN wave model, our data integration engineer, Eneli Toodu created a Github notebook with a guide to using this data. In this guide, you will find instructions on creating visualizations, forecasts, and more with SWAN data from locations like Miami and Houston. We are excited to finally provide access to this premier, high resolution model and are looking forward to the endless possibilities of its applications.


Many of the datasets made available through the Planet OS Datahub have been at the request of our users. For those who require a consolidated, easy to use, resource for accessing large and complex material that the datahub does not already offer, please reach out to the team and we will work toward bringing it onboard. For more information check out the Planet OS Datahub.