Abstract
t Individual precautionary behaviour in response to flooding can considerably reduce flood impacts. Therefore, understanding its drivers and temporal dynamics is of high interest for risk management and communication. Previous studies are mostly based on temporally limited
data by using cross-sectional surveys. Here we identified and characterised different types of
trajectories of adaptive behaviour after a flood event. We used panel data, where 227 households
were repeatedly surveyed within 45 months after the flood of June 2013 in Germany about their precautions. To identify robust groups, we applied and compared two clustering methods: latent class growth analysis(LCGA) and k-means based cluster analysisfor panel data (kmlShape). Three different groups were consistent across the two methods and showed different dynamic adaptive behaviour over the survey period: a ‘high standard’ (35 % of the sample), a ‘high performer’ (37 %) and a ‘low adaptive’ (28 %) group. The high standard group was characterised by a significantly higher protection motivation and flood experience in comparison to the other groups. The high performer group showed the largest increase in implemented precautionary measures after the flood, but also expressed a general fatalistic attitude towards floods. The low adaptive group trusted their community significantly more in managing floods and reported little
access to information and support. The results indicate that tailored risk communication and funding schemes might be needed to support low adaptive types of flood-prone residents. They also present a starting point for the implementation of empirically based, heterogeneous
adaptation behaviour in socio-hydrological models.
data by using cross-sectional surveys. Here we identified and characterised different types of
trajectories of adaptive behaviour after a flood event. We used panel data, where 227 households
were repeatedly surveyed within 45 months after the flood of June 2013 in Germany about their precautions. To identify robust groups, we applied and compared two clustering methods: latent class growth analysis(LCGA) and k-means based cluster analysisfor panel data (kmlShape). Three different groups were consistent across the two methods and showed different dynamic adaptive behaviour over the survey period: a ‘high standard’ (35 % of the sample), a ‘high performer’ (37 %) and a ‘low adaptive’ (28 %) group. The high standard group was characterised by a significantly higher protection motivation and flood experience in comparison to the other groups. The high performer group showed the largest increase in implemented precautionary measures after the flood, but also expressed a general fatalistic attitude towards floods. The low adaptive group trusted their community significantly more in managing floods and reported little
access to information and support. The results indicate that tailored risk communication and funding schemes might be needed to support low adaptive types of flood-prone residents. They also present a starting point for the implementation of empirically based, heterogeneous
adaptation behaviour in socio-hydrological models.
| Original language | English |
|---|---|
| Article number | 103835 |
| Number of pages | 12 |
| Journal | International Journal of Disaster Risk Reduction |
| Volume | 94 |
| Early online date | 23 Jul 2023 |
| DOIs | |
| Publication status | Published - 1 Aug 2023 |