Despite technological advances, marine weather forecasts are still difficult for researchers and scientists to get right. As NOAA’s National Weather Service division explains, “Weather forecasting is not a perfect science and conditions can change rapidly and unexpectedly...Marine forecasting is made much more difficult than forecasting ashore due to a lack of available observations. “
Anyone who’s ever been caught in the rain knows that weather forecasting is never a perfect science, even on land. But, marine forecasting is made more complicated by lack of data, limited access, and reporting inconsistencies. Here’s why marine weather forecasts are often inaccurate and what we can do to improve them.
Why are marine weather forecasts inaccurate?
Marine weather forecasting must overcome the same obstacles as on-shore meteorology, and then some. In the US, on-shore weather predictions suffer from NOAA’s outdated computers as well as decentralized or siloed data collection (as compared to the ECMWF). These issues can sometimes impact marine weather forecasts, too.
But, the biggest issues in accurately predicting marine weather come down to data: we don’t have enough of it, and the data we do have is often unreliable. Why is it so hard to get good data for marine weather forecasting?
The ocean is vast – and filled with obstacles that make it difficult to collect data. Accessibility is complicated by both finding sufficient observation sources and access the data.
Ideally, networks of marine weather buoys would collect accurate data -– and they do, near the coast in water depths that are less than a few hundred meters. But, away from the coast and in deeper water, these networks were previously difficult to set up and monitor. As a result, forecasters were relying on a relatively narrow set of data collected from the same shipping routes over and over again, measuring the same regions of the ocean. Most weather systems, however, originate in remote areas like the Southern Ocean, where ocean observing systems have less access.
Likewise, the size of our oceans makes it very difficult to collect data at scale. The surface area of our oceans approaches 140 million sq miles. Most parts are too remote for wide-band communications. Additionally, saltwater, violent storms, and breaking waves are not electronics-friendly elements. Some companies have overcome these challenges by creating a map of all buoys and retrieving data using an API.
Quantity of data
The inability to access large swaths of ocean territory makes it difficult to get enough data to make accurate predictions. Meteorologists around the world are benefitting from big data. The Korean Meteorological Association, for instance, has increased the agency’s data storage capacity by nearly 1000% to 9.3 petabytes.
As the National Weather Service explains, “Observations are required for NWS computer forecast models and by the forecaster to provide value-added decision making to the computer model output. Observations also serve to alert the forecaster when NWS forecasts do not agree with the actual conditions and therefore, a problem exists in the current forecast process.”
Where weather forecasters on land benefit from thousands of observations – from weather stations, satellite data, weather balloons, and even commercial airplanes – marine forecasters may only be able to draw on one or two observations in a given area.
Quality of data
Not only do we not have enough data – but the data that most scientists use to forecast marine conditions isn’t always the best quality. Weather information from the open ocean comes from a combination of visual observations made by ship crew and satellite-based proxy measurements. These sources have limited accuracy, and can only record and provide data irregular space and time intervals. Even satellite data over oceans is less precise than land-based data. And ship crew observations tend to be anecdotal or flawed at best.
3 ways to improve marine weather forecasts
The good news? Weather models that are in use by NOAA, ECMWF, and others are remarkably accurate. While the technology at NOAA may be a little outdated, the data analytics and computational techniques are not. These models have gotten progressively better since the 1960s. In fact, our numerical weather models are now so good that the skill of our forecasts is entirely limited by the lack of ocean weather data.
The bottom line: if we can improve data collection, we can get more accurate weather forecasts. Here are some ways we can get better marine weather data.
Diversify data sources
To target the issue of lack of quantity of data, one strategy is to deviate from the practice of setting up single “exquisite” sensor systems in favor of distributed networks of low-cost nodes. While single sensors do provide extreme accuracy, they are limited to reporting on a single point. A network of sensors increases density and redundancy, thereby generating more data over a given area that forecasts can feed into models for greater accuracy. The more sensors that set up next to each other, the greater the ability for scientists to verify the data and remove potential outliers, instead of trusting a single sensor.
Incorporate wave data
What do waves have to do with the weather? They’re just another source of information – one that commonly gets left out of marine weather forecasts. Waves can be the first source of information in a more comprehensive global numerical weather model that looks at different variables to predict the weather. Ocean waves are an important part of meteorology that affect both open ocean activities and coastal dynamics. Strong waves can indicate high winds on the surface, and ultimately, feeding real-time sensor data into the weather model can lead to better storm predictions.
Include more data sources
Finally, marine weather forecasters could follow the example set on land. In the US, NOAA experts have managed to upgrade their models by shrinking their focus to the neighborhood level, rather than the city level. The denser our data collection, the better outcome.
Marine weather is still relatively unpredictable, but experts are working to help researchers get more, better data. Learn more about Sofar’s API, which works to provide forecast data and more.