TY  - JOUR
A1  - de Brito, Mariana Madruga
A1  - Kuhlicke, Christian
A1  - Marx, Andreas
T1  - Near-real-time drought impact assessment
BT  - a text mining approach on the 2018/19 drought in Germany
JF  - Environmental research letters
N2  - Contemporary drought impact assessments have been constrained due to data availability, leading to an incomplete representation of impact trends. To address this, we present a novel method for the comprehensive and near-real-time monitoring of drought socio-economic impacts based on media reports. We tested its application using the case of the exceptional 2018/19 German drought. By employing text mining techniques, 4839 impact statements were identified, relating to livestock, agriculture, forestry, fires, recreation, energy and transport sectors. An accuracy of 95.6% was obtained for their automatic classification. Furthermore, high levels of performance in terms of spatial and temporal precision were found when validating our results against independent data (e.g. soil moisture, average precipitation, population interest in droughts, crop yield and forest fire statistics). The findings highlight the applicability of media data for rapidly and accurately monitoring the propagation of drought consequences over time and space. We anticipate our method to be used as a starting point for an impact-based early warning system.
KW  - drought impacts
KW  - Germany
KW  - text analytics
KW  - newspaper
KW  - validation
Y1  - 2020
U6  - https://doi.org/10.1088/1748-9326/aba4ca
SN  - 1748-9326
VL  - 15
IS  - 10
PB  - IOP Publ.
CY  - Bristol
ER  -