Machine learning suggests climate and seasonal definitions should change under global warming

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Extreme and unseasonal temperature and precipitation events have increased worldwide. The greater frequency and variability of floods, heatwaves, and droughts challenge traditional definitions of climate periods as 30-year means. Machine learning (ML) studies, focusing on southern Australia, identified the dominant attributes of these precipitation and temperature events. The attributes are both local and remote climate drivers, amplified by global warming. Their impacts include longer, hotter warm seasons and shorter, drier wet seasons in Australia’s southern Mediterranean climate regions. In contrast, flooding has increased in coastal eastern Australia. The poleward contraction of mid-latitude westerly winds is a readily identifiable contributor. Improvements in climate models are expected to more accurately predict future phases of climate drivers. Because global warming is not uniform across the Earth’s surface, the revised definitions of climate will vary by region. In this work, we chose one clear example that supports the need for re-considering climate periods.

This publication has 16 references indexed in Scilit: