Text mining uncovers the unique dynamics of socio-economic impacts of the 2018–2022 multi-year drought in Germany

Abstract

Droughts often lead to cross-sectoral and interconnected socio-economic impacts, affecting human well-being, ecosystems, and economic development. Extended drought periods, such as the 2018–2022 event in Germany, amplify these impacts due to temporal carry-over effects. Yet, our understanding of drought impact dynamics during increasingly frequent multi-year drought periods is still in its infancy. In this study, we analyse the socio-economic impacts of the 2018–2022 multi-year drought in Germany and compare them to previous single-year events. Leveraging text-mining tools, we derive a dataset covering impacts reported by 260 news outlets on agriculture, forestry, livestock, waterways, aquaculture, fire, and social impacts spanning 2000 to 2022. We introduce the concept of drought impact profiles (DIPs) to describe spatio-temporal patterns of the reported co-occurrences of impacts. We employ a clustering algorithm to detect these DIPs and then use sequence mining and statistical tests to analyse spatio-temporal trends. Our results reveal that the 2018–2022 multi-year drought event had distinct impact patterns compared to prior single-year droughts regarding their spatial extent, impact diversity, and prevalent impact types. For the multi-year drought period, we identify shifts in how impacts have been perceived regionally, especially focusing on legacy and cascading effects on forestry and social activities. Also, we show how regional differences in relevant impacts are controlled by different land-cover types. Our findings enhance the understanding of the dynamic nature of drought impacts, highlighting the potential of text-mining techniques to study drought impact dynamics. The insights gained underscore the need for different strategies in managing multi-year droughts compared to single-year events.