A Holocene flood record of the Lower Rhine

Abstract

Severe floods caused extensive damage and life-loss throughout Europe over the last decades. The magnitude and the short recurrence interval of large events raised questions about the actual safety standards for flood protection. This was reason to raise the 1,250-yr design flood for river dikes in the Netherlands (Waterwet, 2009) from 15,000 to ~16,000 m3s-1. A major problem in calculating the magnitude of the design flood is uncertainty, which mainly originates from limited data availability for predicting the recurrence time of extreme events; a 110-year interval of discharge data presumably poorly represents the distribution of extremes through time. Non-stationarity of the flooding regime further complicates the use of short discharge records for flood frequency analysis, as it is not expected that the distribution and magnitudes of floods is fixed in time. During the Holocene, climate variability and growing human influence have exerted perturbations to the fluvial system, which translates to gradual changes in flood probabilities. In this research available data was increased by harvesting information from sedimentary records from oxbow lake fills of the Lower Rhine. These records were cross-validated with instrumental records, historical records of the last six centuries, and between overlapping sedimentary archives. This resulted in a flood record that stretches back to 8,200 years BP. To arrive at such a record and to use it for flood risk assessments, methodological advances were necessary. First, it was shown how alternative observational information from multiple stations and historical records can be used to extend the discharge record at Lobith. Based on this data it was already demonstrated that floods of the last decades are more rare than previously considered. Moreover, this data clearly indicated multi-decadal non-stationarity of the flooding regime. Next, various types of infilling abandoned channels were studied to target the best sites for retrieving sedimentary flood records; it was concluded that especially oxbow channel fills yield long and subtle flood deposits registration with high preservation potential and recording of flood magnitudes in the coarsest admixed grains in flood beds. Sedimentary characteristics of channel fills and flood beds were correlated with contemporaneous discharge records (since 1770 AD) and major geomorphological changes in the Lower Rhine floodplain. Besides the possibility to use largest events as stratigraphic markers, the correspondence of general fill properties to accumulation rates allowed to refine standard linear age-depth modelling. Discharges of palaeofloods were calculated from the established regression between grain-size characteristics in flood deposits and measured discharges, and hydraulic modelling based on the elevation of slackwater deposits on high terrace levels in the Lower Rhine Valley. It was found that around 4700 years ago, an extreme must have occurred of at least 13,250 m3s-1, larger than any measured discharge. Considering anthropogenic adjustments, this discharge corresponds to at least 14,000 m3s-1 in the present situation, thus reaching similar values as current flood protection levels. Other floods of similar size (millennium floods) occurred around 784 and 1374 AD, and 4500 and 6200 years ago. The 8200-year record also indicates strong temporal variability in the flood regime, especially strong during the Little Ice Age.