A supercomputer is being built to help the UAE’s national forecaster to predict weather patterns more accurately and improve cloud-seeding missions.
The high-performance computer, named Atmosphere, will allow meteorologists to better gauge how and when extreme weather will affect the region.
Developed by Hewlett Packard Enterprise, it will process data up to 600 per cent faster than the current prediction software used by the National Centre for Meteorology.
In only one second, Atmosphere can process millions of calculations and will allow forecasters to issue instant weather warnings to air traffic control and police.
“The system will be finished and operational by end of year,” said Omar Al Yazeedi, director of research, development and training at the NCM.
“[Atmosphere] will allow us to run numerous weather-prediction systems. These systems can calculate and perform complex calculations in order to give us an outlook on how the weather will be in five days, 10 days, even longer lead times if necessary.
“With the previous system we had some limitations. Depending on the computing power you have, you can either do daily operations or research. You cannot do both at the same time.
“With the new system there is enough computing power that you can run multiple systems and carry out scientific research.”
More lead time for weather warnings
He said with increased computing power, the centre can carry out tasks, such as forecasting and delivering weather warnings, much faster.
With previous systems, meteorologists at the centre could finalise their forecast only once all data was in hand, which could take up to three hours to process.
“With the new system it can finish in less than half an hour,” he said.
“That is more than 600 per cent in terms of improvements in enhancement and performance gauges.
“It means we can do multiple data runs in a smaller amount of time, which ultimately leads to a better forecast.”
Noted as the first liquid-cooled system to be deployed for weather prediction in the region, Atmosphere has been customised to deliver advance weather forecasting and overall climate research.
The combined technology will help NCM researchers improve modelling, simulation, artificial intelligence and deep learning capabilities to process complex data, increase accuracy and predict weather events much faster.
“It will help us carry out more complex scientific research and deploy hazard maps in cases of extreme weather events,” Mr Al Yazeedi said.
“The speed of calculation is very important in our field of work. When it comes to extreme weather conditions, every minute counts to save lives and property.
“Prediction is very important and the speed of that prediction is crucial. With this new system it will allow us to give better warnings to other entities like the police and fire services.”
New simulations to allow for better cloud seeding
At present, NCM uses numerous pieces of weather-prediction modelling software, namely the Weather Research and Forecasting (WRF) and Consortium for Small-scale Modelling (Cosmo).
Both systems use physics and dynamics-based models of the atmosphere and oceans to simulate and predict weather conditions.
The new Atmosphere supercomputer will work alongside current systems but will deliver a much faster performance.
Mr Al Yazeedi said the runtime of simulations on WRF will decrease by up to 200 per cent and by up to 300 per cent on Cosmo, speeding up “time-to-insight on weather predictions” significantly.
The supercomputer can also improve current cloud-seeding efforts within the UAE by using AI to predict the best place and time for seeding, which is crucial for such a time-sensitive operation.
“We are proud to collaborate with the National Centre for Meteorology to build its Atmosphere supercomputer,” said Ahmad Alkhallafi, managing director at HPE UAE.
“This is a very important system for any country to have, especially with more extreme weather conditions brought on by climate change.
“As well as forecasting it is critical to the operations of services like air traffic control.”
The NCM runs thousands of simulations to generate real-time weather analysis across a number of national and regional services for the UAE, including aviation safety to guide pilots during take-off and landing at airports as well as to help cities prepare for tropical storms.