Featured Article – AI, Supercomputers … and Storm Dennis
The disruption and damage caused by weather events affect businesses across the UK but having the benefit of early and detailed insights provided a by a new supercomputer system could enable businesses, organisations, agencies, local authorities and the government to plan and minimise disruption where possible, thereby saving billions of pounds per year.
The widely recognised effects of climate change are likely to be responsible for the seemingly more frequent and record-breaking extreme weather events such as storm Dennis here in the UK which brought a record number (594) of flood warnings across England and devastation to many town and businesses.
The increase of a whole degree in temperature since the Industrial Revolution has for example, warmed the atmosphere, making it able to hold more moisture and become unstable. This is likely to mean, at the very least, more storms and flooding in the coming years, and highlights the need for longer advance warnings (Dennis was forecasted 6 days in advance) and more targeted information.
The current Met Office system for weather forecasting utilises 200 billion daily observations from satellites, weather stations and ocean buoys that are interpreted by Cray XC40 supercomputers which are due to reach the end of their life in late 2022. Even though the Cray XC40 is in the top 50 of the world’s most powerful computers, and performs more than 2 million calculations per second for every man, woman and child on the planet, the need for faster, even more accurate, earlier and even more locally-focused weather forecasting requires an investment in new computers to replace the Cray systems for a 10-year period from 2022 to 2032.
The new Met Office supercomputer system, announced against the backdrop of storm Dennis, will cost £1.2 billion (which include hardware and running costs too over a ten-year period) but could, according to the department for Business, Energy and Industrial Strategy (BEIS) deliver £19 worth of economic benefits for every pound spent.
The new system will create a simulated picture of the weather, will divide the globe into grid smaller squares, and will have a massive increase in processing power (six times more powerful) at its disposal which will mean that:
– Four-day weather forecasts will be as accurate today as one-day forecasts were 30 years’ ago.
– With the grid that the simulated picture of the earth’s weather is divided into, the new system will be able to operate (and deliver accurate forecasts) for a resolution as sharp as 100m squares.
– Extreme weather conditions will be spotted earlier.
– Improved daily to seasonal forecasts and longer-term climate projections can be made.
– The effects of a hotter world in the future can be explored, and more detail can be added to projections taking account of factors such as the way nitrogen reacts with the carbon in the air.
– Scenarios such as how the country can make the best use of the land if/when the target of net zero emissions by 2050 is reached can be explored.
– Data from the supercomputer can be used to inform UK government policy as part of leading the global fight against climate change.
The Business and Energy Secretary, Alok Sharma, has highlighted how the new supercomputer looks likely to drive forward innovation and grow world-class skills across supercomputing, data science, as well as machine learning and artificial intelligence. The Met Office’s new supercomputer will, for example, be able to use artificial intelligence and machine learning technology to carry out even more detailed climate model analysis.
Supercomputer Carbon Footprint?
Some commentators have pointed out that the new supercomputers will require a massive amount of electricity to operate them (as the existing ones do), thereby meaning that ironically, they could be contributing to the global warming that is producing the changing weather conditions that they have been introduced to predict.
The Met Office has, therefore, invited potential providers to come up with low-carbon options and it is likely that much of the processing work could be located in countries with easy and abundant sources of clean energy within the European Economic Area e.g. Iceland (geothermal energy) or Norway (hydropower).
Other Weather Forecasting Options For Business
Weather Source and its Snowflake, cloud-based (AWS) data exchange is an example of another accurate forecasting service for businesses that helps them to quantify and manage climate risks. In addition to being fast and providing forecasts projected over 15 days, the service requires only a few hundred gigabytes of new data to be processed per day, therefore, making it a very efficient option.
Looking forward, the socio-economic benefits of the new Met Office system are likely to be many. These will include better forecasting at airports to help the aviation and travel industries, more sophisticated modelling of flooding to help businesses with premises in flood areas with continuity and contingency planning, providing more detailed information to the energy sector which will have knock-on benefits to many businesses, and providing insights and early warnings that are vital for the UK’s transport infrastructure and for business transportation, delivery and distribution services.
In a world where the climate is capable of causing massive damage and disruption on a more regular basis, the investment in even more powerful supercomputer technology and AI by the Met Office could actually bring considerable savings for businesses of all kinds and provide a better basis for government decisions.