New Waterway, can exert a more apparent impact on extreme water levels than sealevel rise. Land subsidence in delta regions influences flood risks worldwide (Syvitski et al., 2009). The rate of land subsidence in the western part of The Netherlands is estimated between 1 and 10 mm/y, depending on soil characteristics. These numbers have the same order of magnitude as the centimetre-scale gradual changes in tidal amplitude that have been observed in the river network between 1981 and 1997. Changes in tidal amplitudes, even on a local scale, influence flood level statistics. Since safetymeasures and flood protection programmes are often based on these flood level statistics, it is important to establish the degree in which engineering measures influence these statistics. Humaninduced water level changes have the same order of magnitude as sea level rise and land subsidence, but develop over much shorter time periods. Also, the spatial variability of changes resulting fromlocal engineering works is comparatively high. Meaningful flood level statistics therefore cannot be derived from historical water level data on a small number of locations, butmay better be obtained froma hydrodynamical model with a fixed channel configuration and bathymetry, covering the complete channel network.
6. Conclusions
Water levels in the Rhine–Meuse river network have beenmeasured since the 1940s. The resulting dataset provides an excellent opportunity to investigatewater level variation in detail, and to link these changes to specific human interventions. Although mean water levels in the area show the same general rising trend as the sea level, there is an additional fluctuation overwhelming the long-term trend that is coherent throughout the channel network. High and low water level statistics feature a large temporal and spatial variability. Combined Mann− Kendall tests reveal that mean water levels increase, while both extreme low and extreme high water levels decrease. The Pettitt-test yields significant change-points in extreme water levels in 1970, 1980 and 1997. Results from a harmonic analysis indicate that tidal amplitudes underwent sudden changes in these years. The changes in tidal motion and in the associated change in extreme water level statistics can be attributed to three isolated engineeringmeasures.Most apparent is the closure of the Haringvliet estuary in 1970. Secondly, linking the Hartel Canal to the OudeMaas at its eastern end caused a regimechange in 1981, creating a retention basin in the network. Lastly, trend break was established when the Hartel Canal became an active link in the river network after its westernmost part was connected to the New Waterway in 1997. Gradual changes in water level are most likely caused by dredging. The abrupt and gradual impacts that engineering works have on high and low water level statistics are of importance for flood risk. The rates of water level change induced by engineering works are similar to or larger than the effect of sea level rise. This warrants more attention to the long-term influence of human engineering on flood risk in tidal river networks.