Forecasting for the future

With climate change dictating more frequent extreme weather conditions, oceanographers are turning to new data providers. 

Whether aboard a large container ship, maintaining offshore wind turbines or setting out to trawl valuable fisheries, seafarers rely on accurate and timely weather forecasts for their vessel’s safety. While the numbers of large sea-going ships lost at sea is falling due to an improved focus on safety management and training, there are some aspects of life at sea that are harder to control: the weather. Indeed, casualty statistics from the IMO reveal that 'bad weather' is a contributing factor in one in five ships losses, making accurate, accessible and timely weather forecasts essential.  

Climate change, however, means the historical records and data points that power existing models are less helpful when it comes to forecasting what comes next. To truly understand the complex systems that drive our planet’s weather, meteorologists, oceanographers and data scientists need ever more data about what’s happening in real-time as well as historical records that show what happened before.  Yet there are huge tracts of the earth where there’s scant data, with vast areas of the oceans, increasingly recognised as the central player of the climate system, considered 'data sparse'.  

Absorbing the changes 

According to the World Meteorological Organisation (WMO), the oceans have 1,100 times the heat capacity and over 100,000 times as much water as the atmosphere, and this is being tested as never before. Experts predict that by 2100, the ocean will have absorbed up to four times more heat than it has in the last 50 years – and that’s if global warming is limited to 2°C.  

The impacts are many, from reduced oxygen, ocean acidification and changing ocean chemistry that will have significant, if not catastrophic, consequences for the ocean food web, marine ecosystems and the people that depend on them as well as the currents and forces that shape our planet’s climate. Already these changes are feeding into more extreme weather events or more unpredictable patterns, including unexpected shifts in the prevailing winds which have clear implications for mariners – many of the ports and terminals in the south east of England, for example, were developed to deal with winds and weather from a predominantly south-westerly direction but there has been a shift towards increased easterly winds. 

The data sparsity of many ocean areas was compounded by the pandemic, as the global network of floats and buoys that capture valuable oceanographic data suffered deployment slowdowns and maintenance outages when research and commercial vessels were curtailed by COVID-19 restrictions.  The National Oceanic & Atmospheric Administration has been part of a project to launch 100 new Argo floats – a global array of about 3,800 floats that measure pressure, temperature and salinity of the upper 2,000 metres of water - across the Atlantic. The data will be used to support ocean, weather and climate research and help “improve data that drive life-saving weather and climate forecasts”, according to a NOAA spokesman.  About 1,000 Argo floats need to be deployed each year to sustain the Global Ocean Observing System offsite link, which has helped collect data that has transformed understanding of the oceans. 

Data gathering is also being crowd-sourced, with the IMO encouraging mariners to participate in the WMO Voluntary Observing Ships (VOS) Scheme. Over 4,000 ships are registered in VOS, submitting more than 2.5 million observations which, in some remote areas such as the polar regions, can be the only data available.  

Understanding the impact 

Taking these huge data sets and integrating them with data from the atmosphere, the terrestrial realm, the cryosphere and biosphere are increasingly important to predict the weather.  This 'Earth System' approach is vital to help with forecasting and understand not just what the weather will do, but what it’s impact will be – a key shift that the WMO is keen to promote as we learn to live and adapt to climate change.  

But it’s not just about the volume of data – it’s about what you do with it and increasingly that means deploying Artificial Intelligence and machine learning. Google’s AI technology, DeepMind, for example, has partnered with the Met Office to apply machine learning to rainfall 'now-casting' (typically highly granular predictions for what’s going to happen in the next two hours). DeepMind uses radar data and generative modelling to make detailed predictions based on past weather.  

Another AI-powered forecaster, Boston-based start-up Salient, which takes a “machine-learning- first” approach, has developed a model that links precipitation rates to ocean salinity. Newly appointed CEO Matt Stein says the system consistently outperforms the large government agencies and private forecasters.  

Matt Stein, CEO of forecaster Salient

"Traditionally, weather forecasts have been focused on the 10-day outlook,” said Stein at the time of his appointment, highlighting that this “proprietary technology now makes it possible to forecast in the timeframes in which businesses think—monthly, quarterly and beyond."  

As climate change continues to wreak havoc across our world, these new models and the forecasts they produce – from 'now-casting' to longer term seasonal outlooks – will be essential for decision-makers across the marine professions.   

Catch-up on recent webinars and events about alternative fuels and decarbonisations on IMarEST TV (for IMarEST members only). 

Amy McLellan is a journalist and author. She was previously editor of Energy Day. Twitter  @AmyMcLellan2