AI­ powered weather forecasting from Google

Context: AI is revolutionising weather forecasting by using machine learning to process large amounts of historical weather data and provide more accurate 10-day forecasts in less than a minute. This technology is being developed by Google’s GraphCast program in their DeepMind AI research lab.

Details:

• Google DeepMind has developed an AI model that can predict extreme weather events caused by climate change, allowing us to better prepare for and mitigate natural disasters, potentially saving lives.
• The model has achieved a verification rate of 90% and outperforms traditional weather prediction technologies in terms of accuracy. 
• In certain cases, GraphCast has been found to be 99.7% more accurate than the leading system. 
• This open-source tool has the ability to identify extreme weather events and has the potential to improve its accuracy by incorporating updated data. 
• The GraphCast weather prediction program utilises the two most recent states of Earth’s weather, including variables from the present and six hours earlier, to forecast the weather conditions for the next six hours to 10 days.

Main Aspects:

• • GraphCast, developed by Google DeepMind, provides meteorologists with early and accurate warnings about extreme temperatures and cyclone paths. 
    • In a recent example, GraphCast accurately predicted Hurricane Lee’s landfall in Nova Scotia nine days before it happened, while traditional models only predicted it six days in advance.
  • According to Pushmeet Kohli, the vice president of research at Google DeepMind, weather prediction has been a difficult problem that has been studied by humanity for a considerable amount of time. 
    • Given the recent impact of climate change, this problem has become exceptionally significant.
  • Machine learning: GraphCast is a system that uses machine learning to quickly calculate weather predictions. 
  • Neural networks :It employs graph neural networks to map the Earth’s surface into a million grid points and predicts various weather conditions such as temperature, wind speed and direction, mean sea-level pressure, and humidity. 
    • The neural network analyses patterns in the data to make predictions for each of these grid points.
  • Disadvantages: GraphCast is not flawless and is not as advanced as traditional weather forecasting models in certain aspects, like predicting precipitation. 
    • Therefore, meteorologists will need to combine traditional models with machine-learning models in order to improve the accuracy of their predictions.

• Advantages: An AI with advanced weather forecasting capabilities would offer numerous benefits. 

• The goal is to minimise the negative effects of unpredictable weather on agriculture, tourism, and transportation. 

• This includes implementing safety measures to protect the population from dangerous climate events and adjusting environmental policies based on long-term climate trends. 

• With tools like GraphCast, we are making progress in understanding our environment and being able to predict weather changes more effectively.

Conclusion:

• DeepMind has developed an artificial intelligence system that greatly improves the accuracy of weather forecasting using deep learning and convolutional neural networks. 
• This technology has the potential to enhance economic decision-making, protect people from climate-related risks, and monitor climate change, leading to a more secure and controlled future in meteorology.

Source: TOI

Download as PDF