From dike history to reinforcement practice

Throughout the centuries most river dikes in the Netherlands have been reinforced multiple times, which has resulted in highly variable dike interiors. Dike interior material and its variability is an important factor in dike safety calculations. Archeological observations can increase the data availability of historical dikes and improve simulations of dike interior variability.

As an  All-Risk researcher at the Utrecht University , most of my work so far has focused on simulating groundwater flow through the natural subsoil that is naturally composed of several layers below the dikes. However, when a colleague showed me an image of his archaeological investigation of a dike related to a reinforcement project, I was just as surprised as he was: Several distinct layers of different material types were visible. The many layers of the dike interior shouldn't have come as a surprise, as I knew many river dikes in the Netherlands have a long history. Throughout their existence, the river dikes were heightened and widened at many occasions. At each enforcement, new soil was added consisting of (slightly) different material, making the current river dike very variable (as can be seen on the images below).

To incorporate variable dike interiors in the dike safety assessment, I developed DETRIS: Dike Erection Tessellation using Regionally Inherited Statistics. The algorithm uses statistics on historically used material types, reinforcement layer thicknesses and dike surface steepness to construct hypothetical dike cross-sections. Here is a simulation of this DETRIS dike compared with the real construction history of the Bemmelsedijk. The layers in the DETRIS dike pile up following a similar sequence. For nearby river dikes, it might also be possible to simulate the dike interior by looking at the construction history at locations where we do not have an archaeological dike observation.

The DETRIS algorithm allows to simulate the many layers of the dike interior based on archaeological dike properties. This opens up several opportunities:

• Groundwater simulations can better estimate the evolution of groundwater pressures during high river water levels by considering not only the layers below the dike but also the dike interior.
• Dike slope stability problems can consider non-homogeneous dike interiors and by creating multiple simulations of variable dike interiors, a probabilistic analysis of the dike slope stability can be performed.
• Validating the DETRIS simulations with hydrological, geophysical or geological data can improve the reliability of the simulated DETRIS-dikes.

While the algorithm can accurately recreate the material composition in observed historic dikes, two problems remain:

• The variation in material types and dike history in the Netherlands is high on a small spatial scale, so dikes located 1 km apart may look very different. With the algorithm being very dependent on the observations that is it based on, making predictions of locations where detailed archaeological observations are not close enough lowers the precision of the algorithm.
• Variability in the dike interior is not only a result of different reinforcement periods. Dike breaches, and the rapid filling of the breach with any material available, can also cause interior dike variability. 

Technical managers working for the water authorities could already use the DETRIS algorithm to better calculate dike safety. However, the best way to further improve the algorithm is by adding more observations on the river dike interior to the database. These archaeological observations are necessary to make a better estimation of the characteristics at a dike location, which, in turn, improve the DETRIS predictions of nearby dikes. I would encourage any cooperation between specialists in the fields of dike safety and dike history, as they proved to be much more intertwined than previously thought. This cooperation, may in the end to much smaller uncertainties in dike safety estimates and thus in safer river dikes all across the Netherlands.

•  A fast geotechnical survey of dike cross-sections  (in Dutch)
•  Assessing lithological uncertainty in dikes: Simulating construction history and its implications for flood safety assessment ; van Woerkom et al. (scientific article).