An international study team (POSTECH) has revealed a major change in the properties of worldwide daily precipitation using a deep learning approach. The findings showed that, since 2015, the daily precipitation pattern had clearly deviated from natural variability on more than 50 per cent of all days, which was driven by human-induced global warming. The researchers deployed explainable artificial intelligence to show that changes in daily precipitation variations were gradually intensifying upon weather timescales. Conventional linear statistical techniques employed in earlier studies on climate change detection had difficulties identifying non-linear effects, such as the increased variability in daily precipitation, but deep learning got over these restrictions by using non-linear activation functions.
Add to Favourites
To login click here