Abstract
At river bifurcations, discharge is distributed over downstream branches, thereby also distributing flood risk.
An accurate estimate of the discharge distribution is essential for flood risk management. In our research, we
analyze the water balance and discharge distribution at two major bifurcations of the Rhine river in the Netherlands. The Dutch Rhine branches are highly engineered, and human intervention will continue in the future.
For each of these branches, a 30-year record of discharge and water level measurements is available. Currently
operational rating curves show non-closing water balances at the bifurcations, with errors of up to 10%. This
indicates large uncertainties in these rating curves. In our study, we construct multi-stage rating curves using Bayesian inference and Markov chain Monte Carlo simulations. We explore two methods to increase rating
curve accuracy. First, we will take explicitly into account the constraints of the river bifurcations, namely a closing water balance, in the construction of the Bayesian rating curves. Secondly, we will consider water levels
along downstream branches as an indicator for discharge measurement inaccuracy. The results show that both
methods are able to strongly reduce the water balance error, particularly at very high discharges. However,
this comes at the expense of a wider uncertainty range in the rating curves. Still, the new rating curves may be
more accurate due to the reduced water balance error. Using the rating curves, we quantify the 90% confidence
interval of the discharge distribution at both bifurcations at around 10%, a value which is fairly independent
of upstream discharge. At design conditions, this amount of uncertainty could result in uncertainties in water
levels of up to 0.8m. We conclude that it is essential to regard the entire river system with its bifurcations for
flood risk management and for future planning of human interventions in the system.
An accurate estimate of the discharge distribution is essential for flood risk management. In our research, we
analyze the water balance and discharge distribution at two major bifurcations of the Rhine river in the Netherlands. The Dutch Rhine branches are highly engineered, and human intervention will continue in the future.
For each of these branches, a 30-year record of discharge and water level measurements is available. Currently
operational rating curves show non-closing water balances at the bifurcations, with errors of up to 10%. This
indicates large uncertainties in these rating curves. In our study, we construct multi-stage rating curves using Bayesian inference and Markov chain Monte Carlo simulations. We explore two methods to increase rating
curve accuracy. First, we will take explicitly into account the constraints of the river bifurcations, namely a closing water balance, in the construction of the Bayesian rating curves. Secondly, we will consider water levels
along downstream branches as an indicator for discharge measurement inaccuracy. The results show that both
methods are able to strongly reduce the water balance error, particularly at very high discharges. However,
this comes at the expense of a wider uncertainty range in the rating curves. Still, the new rating curves may be
more accurate due to the reduced water balance error. Using the rating curves, we quantify the 90% confidence
interval of the discharge distribution at both bifurcations at around 10%, a value which is fairly independent
of upstream discharge. At design conditions, this amount of uncertainty could result in uncertainties in water
levels of up to 0.8m. We conclude that it is essential to regard the entire river system with its bifurcations for
flood risk management and for future planning of human interventions in the system.
| Original language | English |
|---|---|
| Pages | 4 |
| Number of pages | 1 |
| Publication status | Published - 10 Aug 2021 |
| Event | 8th International Conference on Flood Management 2021: Lowering Risk by Increasing Resilience - Virtual Event, Iowa City, United States Duration: 9 Aug 2021 → 11 Aug 2021 Conference number: 8 |
Conference
| Conference | 8th International Conference on Flood Management 2021 |
|---|---|
| Abbreviated title | ICFM8 |
| Country/Territory | United States |
| City | Iowa City |
| Period | 9/08/21 → 11/08/21 |