Impacts of Climate Extremes from Mining of Online Texts

Climate extremes have multifarious detrimental impacts on human activities and ecosystems.

Gaining a detailed understanding of these impacts is essential for disaster risk reduction and for building resilience to extremes in a changing climate.

However, current freely-accessible natural hazard databases present limitations in completeness, updateability, validation against climate data and indirect impact information.

This hinders scientific and practical progress.In ICE-MOT, I aim to build a state-of-the-art impact database for extreme climate events, which overcomes the above key limitations.

I will specifically combine text-mining of freely available online sources, with the use of state-of-the-art climate data.

To ensure feasibility, I will initially focus on English-language texts and on wintertime cold spells in North America and windstorms and heavy precipitation in Europe extreme events which I have studied extensively in my ongoing ERC project.ICE-MOT builds upon the database of climate extremes developed within my ongoing ERC project.

It further leverages the experience of my research group in data-driven and machine learning analyses for climate science.

I will use this interdisciplinary knowledge base to provide standardised, complete and automatically updateable spatio-temporal impact information, including indirect and/or cascading impacts, and quantify the climate conditions associated with the recorded impacts.

Moreover, the databases automated data extraction and processing pipeline will make it easily scalable to multiple regions and climate extremes.This effort is timely: the recent EU strategy on adaptation to climate change explicitly seeks to gather more and better data on climate-related risks and losses as a key adaptation tool.

Moreover, the climate extremes data gathered in my ongoing ERC project provides a perfect basis to test the innovative idea underlying the ICE-MOT database, an opportunity which should be rapidly exploited.