- Home
- The Pulse
- The Pulse: Articles
- Charting Intelligent Seas: The Technological Architecture Behind Weather Navigation
Charting intelligent seas: The technological architecture behind Weather Navigation
Global shipping moves around 90 percent of world trade, yet its routes remain remarkably inefficient. Every day, thousands of vessels cross oceans without fully accounting for the shifting interplay of wind, waves and currents that governs their energy use. The consequences are significant: excess fuel consumption, higher operational costs and avoidable carbon emissions.
Weather Navigation seeks to change that. This advanced algorithmic system uses real-time environmental data and large-scale computation to chart routes that adapt to constantly changing marine conditions. While it has been affectionately nicknamed “the Google Maps of the sea,” in reality, it is far more complex. Beneath the surface lies a rigorous mathematical framework that fuses stochastic modeling, optimization theory and high-frequency data integration—all aimed at making maritime transport both smarter and more sustainable.
Developed by a team of data scientists, mathematicians, computer scientists and business management experts with a passion for sustainability—led by IE University professor David Gomez-Ullate and including professor Daniel Precioso Garcelán—Weather Navigation exemplifies how artificial intelligence and environmental science can converge to transform a traditional industry.
But how does the technological infrastructure that Weather Navigation uses work in practice? Here, we speak to Dr. Precioso to explore the technological backbone that makes Weather Navigation’s approach to route optimization possible.
How does Weather Navigation aggregate weather data into a single routing system?
The weather-related data that Weather Navigation draws from is publicly and freely available. It comes from enormous and established repositories like the National Oceanic and Atmospheric Administration (NOAA); the European Union's Earth observation program, Copernicus; and the HYCOM consortium, a multi-institutional effort sponsored by the National Ocean Partnership Program.
The marine service of Copernicus, for example, provides free, regular and authoritative information on the state of what it calls the blue (physical), white (sea ice) and green (biogeochemical) ocean, both regionally and globally. “Anyone can visit the Copernicus website and take a look at the data we’re using,” says Dr. Precioso. “This source, and the others we use, offer detailed information for a variety of metrics, including ocean currents, wind patterns and wave behavior.” Weather Navigation, he explains, brings together data from various platforms in one place: “We download the data, store it in the cloud and ensure ships can access this information anywhere.”
The biggest challenge lies in the fact that the different forecasts aren’t always consistent—anyone who’s ever used multiple apps to anticipate the weather for a Saturday barbecue will know the feeling! The repositories adopt different models and the weather itself is subject to change. This is a particularly interesting challenge, according to Dr. Precioso, and the team is looking into using robust optimization in order to address it.
Robust optimization is a mathematical approach to solving real-world optimization problems (like weather routing) that have uncertain parameters (like varying weather predictions). Rather than using a single set of values, robust optimization analyzes a range of possible values, known as an uncertainty set, and then finds a solution that is feasible for all possible scenarios in that set. It’s about finding coherence amid variability, and it will help Weather Navigation develop routes that are resilient to forecasting discrepancies and errors.
“We know that weather forecasts are never cast in stone,” Dr. Precioso explains, “and that our solution needs to be inherently flexible. At the moment, we’re doing everything we can to recommend the best possible route—the route that has the least chance of failing—within the parameters we can access.”
How often does the system download and process changing weather forecasts?
Weather Navigation monitors its sources on a continuous 24-hour cycle. “We have an application that pays attention to when NOAA, Copernicus and HYCOM update their data,” he explains. “The moment they release a new forecast, our application downloads it.” In the case of Copernicus, this typically happens around midnight, and the entire download process takes less than an hour.
This efficiency means the system can begin recalculating routes with fresh data in the early morning hours, well before most shipping operators require updated route recommendations.
“Every time we download new weather data, the service runs again and rechecks its suggested route,” Dr. Precioso explains. “Captains then receive daily route suggestions with up-to-date forecasts and course corrections if necessary.”
This continuously updated process is fundamentally better than traditional weather routing systems, which typically involve time-consuming emails back and forth between a ship and a shore-based consultant. “With Weather Navigation, if the forecast changes—which is inevitable—the system automatically recalculates and sends updated recommendations the next morning,” he continues. “And if unexpected conditions emerge mid-voyage, the captain can request a newly optimized route at any time, and receive results within five to ten minutes.”
The system stores this data on the cloud and only retains the most up-to-date 10-day window. Yesterday’s data, after all, is no longer needed for any future forecasting. This approach helps Weather Navigation to manage its storage requirements, while ensuring that it always works with the most current information. And the 10-day window is strategically chosen. Many shipping routes take between a few days to two weeks to complete, so 10 days provides enough data for ships to plan in advance while staying within a reliable forecasting horizon.
“It’s worth bearing in mind that Weather Navigation functions as a decision-support tool, not a control system,” Dr. Precioso emphasizes. “It’s suggesting the best option, but the captain retains full authority on which route is best. With this data on hand, however, he’s able to make a totally informed call.”
What kind of computational power does it take to recommend optimized routes?
The computational power required to deploy and use Weather Navigation is perhaps one of the most impressive things about it. Thanks to faster processors and advanced cloud infrastructure, computation that once may have taken hours can now be done in minutes.
For ships at sea, the data requirements are minimal. The heavy computational work happens on Weather Navigation’s cloud servers. Ships only need to send basic route requests, which are similar in size to a WhatsApp message, and they receive optimized routes in return. “Ships can operate this service with very little internet connection,” says Dr. Precioso. “This is critical given the often limited bandwidth available to vessels in the middle of the ocean.”
In order to understand why this is so impressive, it’s worth diving into Weather Navigation's computational approach. Essentially, in its development, Weather Navigation has opted to compromise between two different philosophies, leading to what the team describes as a “two-stage pipeline.”
The first stage uses a stochastic algorithm. This is a computational method that incorporates randomness and probability; it’s an algorithm that doesn’t produce the same output every time. This randomness initially concerned the mathematicians on the team who were accustomed to looking for reliable and consistent answers. But it serves a crucial purpose in its ability to quickly generate safe routes that avoid obstacles like coastlines, islands and dangerous weather zones.
“This approach ensures that the boat isn’t going to hit land,” Dr. Precioso explains. While this might sound simple, from a software perspective, collision avoidance in marine environments is incredibly complex and computationally demanding. The stochastic approach handles this efficiently, producing a route that is safe and reasonably optimized.
The second stage then refines this initial route using deterministic mathematical optimization. These algorithms, which leverage concepts like gradient descent and variational optimization, smooth the route and find the best possible route given the constraints.
This staged approach works well in terms of computational resources, however. It splits the optimization process into an initial phase that is more technically demanding and a secondary phase that is technically lighter, where algorithms can be fine-tuned as new weather data becomes available.
In the long term, the intention is for Weather Navigation to serve thousands of vessels simultaneously, a prospect in which Dr. Precioso has absolute faith. “The cloud-based infrastructure means that computational capacity can expand as demand grows, without requiring expensive hardware installations on every vessel or complex training.”
How does Weather Navigation differ from other maritime routing providers in the market?
Weather Navigation is not the only marine route optimization solution on the market—but it is unique in its approach.
“Most of the other solutions available come from the shipping world,” Dr. Precioso explains. “Their biggest strength is that they know ships very well; they have excellent ship modeling.” These competitors invest significant resources in creating detailed vessel performance models that can predict precisely how a specific ship—with its particular hull shape, length, weight, center of gravity and cargo load—will behave under certain conditions.
Of course, this information is enormously useful. When a shipping company knows exactly how its vessel performs, it can make informed operational decisions and gain accurate insights into fuel consumption and speed predictions. And it’s certainly data that Dr. Precioso and his team intend to incorporate in the algorithm’s future iterations. But for now, Weather Navigation is adopting an alternative approach.
“We’re focusing on the optimization part, on uncovering the very best route. We’re researching state-of-the-art mathematical models and using them to inform the most advanced route-finding algorithms,” he explains.
This means that while other solutions might tell you precisely how much fuel your specific vessel will consume on a given route, Weather Navigation will find you a fundamentally better route: one that leverages favorable currents, avoids harsh weather, optimizes timing, and saves fuel and reduces emissions in the process. “Even with perfect knowledge of your ship, if you’re not optimizing the route properly, you’re leaving savings on the table,” he adds.
Looking ahead, Weather Navigation is likely to combine dedicated ship profiles and performance metrics with sophisticated mathematical optimization. “But our immediate priorities are to combine the best mathematical models with robust optimization and total flexibility so that every ship can find the ideal route and adapt decision-making quickly.” Dr. Precioso explains. “And I think that is our biggest difference.”
What comes next?
Weather Navigation is a major technological achievement. But it’s important to recognize that it is still in its developmental stages. The system currently operates as a sophisticated prototype, tested with select shipping companies and refined through ongoing research partnerships—there is more to come.
Several aspects of the technology will inevitably evolve. Vessel-specific modeling, robust optimization algorithms and real-world validation with large commercial fleets are all still to come, as is the challenge of building trust within a traditionally conservative industry. And the computational architecture itself may shift as the system scales from serving dozens of routes, shipping companies and ships to potentially thousands of each.
What is clear is that Weather Navigation has established a sophisticated algorithm that meets the shipping industry’s urgent need for decarbonization solutions. As the technology matures, there is little doubt that it will continue to evolve, improve and ultimately help chart a more sustainable course for global maritime transport.