Strijker et al. (2025) The dynamics of peak head responses at Dutch canal dikes and the impact of climate change

Identification

• Journal: Natural hazards and earth system sciences
• Year: 2025
• Date: 2025-09-12
• Authors: Bart Strijker, Matthijs Kok
• DOI: 10.5194/nhess-25-3355-2025

Research Groups

• Hydraulic Engineering, Delft University of Technology, Delft, the Netherlands
• Risk and disaster management Unit, HKV Consultants, Lelystad, the Netherlands

Short Summary

This study investigated the dynamics of peak hydraulic heads in Dutch canal dikes at a national scale using non-linear time series models calibrated on extensive observation data. It found that climate change will significantly alter the frequency of extreme peak heads, with projections indicating occurrences between 3 times less and 8 times more frequently by 2100, depending on the climate scenario and dike characteristics.

Objective

• To assess the dynamics of peak hydraulic heads in Dutch canal dikes on a national level, caused by heavy rainfall events, by analysing the variation in head responses and head statistics.
• To understand the reasons for differences in head dynamics by relating these variations to the physical properties of the dikes.
• To quantify the potential impact of climate change on head statistics, indicating how flood risks in Dutch polders are expected to change in the future.

Study Configuration

• Spatial Scale: National level in the Netherlands, focusing on canal dikes primarily in the western and northern polder regions. Data collected from 48 monitoring sites across seven regional water authorities.
• Temporal Scale:
  • Head observations: 2006–2023 (daily-mean values), with individual time series lengths varying from 3 to 9 years (average 5 years).
  • Simulations: 30-year time series for current and future climate scenarios.
  • Future climate scenarios: Two time horizons (2050 and 2100) with two greenhouse gas emission pathways (SSP1-2.6 and SSP5-8.5) and two regional climate responses (wet-trending and dry-trending).

Methodology and Data

• Models used:
  • Time series models, specifically Transfer Function Noise (TFN) models implemented in the Pastas (version 1.6.0) Python package.
  • Four model structures evaluated: Linear-Exponential, Linear-Gamma, Flex model (non-linear), and TARSO (Threshold Autoregressive Self-exciting Open-loop) model (non-linear). The TARSO model was selected as the best performer.
  • Generalized Pareto Distribution (GPD), specifically an exponential distribution, fitted to selected peak heads for extreme value analysis.
  • k-means clustering algorithm for identifying dike clusters based on peak head coincidence.
  • Statistical tests: Wald test, Wald chi-squared test, and Kruskal–Wallis test for assessing relationships between variables.
• Data sources:
  • Head observations: 108 daily-mean hydraulic head time series from 48 monitoring sites in Dutch canal dikes, collected from seven regional water authorities.
  • Precipitation and evaporation:
    • Historic local weather: KNMI radar-derived precipitation and inverse distance weighting (IDW) interpolated potential evaporation from KNMI ground stations.
    • Long-term climate scenarios: Nine 30-year precipitation and evaporation time series from KNMI (station Aalsmeer) representing current and future climate conditions.
  • Physical dike characteristics: Subsurface material (from borehole descriptions, cone penetration tests, or GeoTOP 3D model), head difference (canal-polder water levels), dike slope (derived from Actueel Hoogtebestand Nederland elevation map), and equivalent drainage length.

Main Results

• The non-linear TARSO model was identified as the best-performing model structure (average R² of 0.74), accurately capturing the non-linear head dynamics in Dutch canal dikes. 35 out of 48 monitoring sites (73%) yielded reliable models.
• TARSO models consistently showed that the lower regime has longer response times (typically 100–600 days) and lower peak block responses compared to the upper regime (typically 2–50 days).
• Four distinct dike clusters were identified based on the coincidence of peak heads, primarily differentiated by the response time of their upper regime. These clusters did not exhibit a clear spatial pattern.
• Subsurface material (clayey or peaty) was found to be a statistically significant factor influencing dike clusters (p-value = 0.03). Dikes with shorter equivalent drainage lengths (less than 20 m) were associated with clusters having smaller response times.
• The median decimate height (increase in head level for a 10-fold increase in return period) across the canal dikes was 0.15 m, ranging from 0.05 m to 0.50 m. Lower decimate heights were associated with smaller peak block responses and shorter response times in the upper regime.
• By 2100, extreme peak heads are projected to occur between 3 times less and 8 times more frequently, depending on the climate scenario and the specific canal dike. Dikes with longer response times in the lower regime appeared less impacted by climate change.

Contributions

• Provided the first national-scale assessment of peak hydraulic head dynamics in Dutch canal dikes, utilizing a unique and extensive dataset of head observations.
• Identified and validated the non-linear TARSO model as the most suitable approach for simulating complex head dynamics in these dikes.
• Quantified the spatial variability of head responses through a novel clustering approach based on peak head coincidence, revealing distinct response groups.
• Established relationships between variations in head responses and physical dike characteristics, highlighting the influence of subsurface material and dike geometry.
• Quantified the projected impact of climate change on extreme peak head statistics and frequencies for Dutch canal dikes, offering critical insights for future flood risk management.
• Introduced and analyzed the concept of decimate height in canal dike head statistics, demonstrating its relevance for dike safety assessments.
• Made the collected head observation data and model scripts publicly available, fostering reproducibility and further research in the field.

Funding

• Stichting Toegepast Onderzoek Waterbeheer (STOWA)
• Rijkswaterstaat

Citation

@article{Strijker2025dynamics,
  author = {Strijker, Bart and Kok, Matthijs},
  title = {The dynamics of peak head responses at Dutch canal dikes and the impact of climate change},
  journal = {Natural hazards and earth system sciences},
  year = {2025},
  doi = {10.5194/nhess-25-3355-2025},
  url = {https://doi.org/10.5194/nhess-25-3355-2025}
}