An assessment of uncertainty in flood hydrograph features, e.g., peak discharge and flood volume due to variability in the rainfall-runoff model (HEC-HMS) parameters and rainfall characteristics, e.g., depth and duration, is conducted. Flood hydrographs are generated using a rain pattern generator (RPG) and HEC-HMS models through Monte Carlo simulation considering uncertainty in stochastic variables. The uncertainties in HEC-HMS parameters (e.g., loss, base flow, and unit hydrograph) are estimated using their probability distribution functions. The flood events are obtained by simulating runoff for rainfall events using the generated model parameters. The uncertainties due to rainfall and model parameters on generated flood hydrographs are evaluated using the relative coefficient of variation (RCV). The results reveal a higher RCV index for flood volume (RCV = 153) than peak discharge (RCV = 116) for a 12-hr rainfall duration. The average relative RCV (ARRCV) index computed for hydrological component (e.g., base flow, loss, or unit hydrograph) indicates the highest impact of rainfall depth on flood volume and peak. The results indicate that rainfall depth is the main source of uncertainty of flood peak and volume.
Reliable estimation of flood characteristics is essential for flood mitigation planning and designing of urban hydraulic structures [
The prediction of floods using rainfall-runoff models are associated with uncertainty due to the uncertainty in input variables (i.e., rainfall), model parameters (i.e., loss), and model structure [
Besides, inaccuracy in rainfall-runoff model parameters, their scale, and other associated errors impart uncertainty in flood hydrographs [
To the best knowledge of the current study, the uncertainty analysis of flood hydrograph is conducted using various considerations. The flood hydrographs of Jamishan dam catchment located in Iran are generated using the rain pattern generator (RPG), which is a stochastic rainfall model and a physically based lumped hydrological model known as HEC-HMS. The uncertainties in rainfall duration and depth, as well as HEC-HMS parameters, are considered for the estimation of flood hydrographs. The Monte Carlo simulation is used for the evaluation of uncertainties of input variables and model parameters. The selection of this method is due to its remarkable advantages such as flexible constrained simulation and the potential to account the dependence between input variables. Finally, propagation of uncertainty of input variables/model parameters to the flood hydrographs is studied.
The Jamishan basin, located in Kermanshah province, Iran, with an area of 524.07 km2, is used as the case study area in the present study (Figure
(a) The Jamishan River basin, and (b) the digital elevation model of the basin.
This study assesses the uncertainty associated with flood hydrograph features namely peak discharge and flood volume due to variability of the rainfall-runoff model (HEC-HMS) parameters and rainfall characteristics namely rainfall depth, duration, and pattern. The rainfall events are generated using a stochastic model named rainfall pattern generator (RPG), introduced by Sharafati and Zahabiyoun [
The RPG stochastic rainfall model generates random rainfall events via MCS sampling procedure. The model characterizes a rainfall event in terms of rainfall depth, duration, and rainfall pattern. It assumes that the rainfall pattern is dependent on rainfall depth and duration. To provide the PDFs required for generating the rainfall events, observed data are divided into several depth classes. Subsequently, rainfall events for each depth class are distributed to some duration classes, and the rainfall events in each duration class are categorized into rainfall pattern classes. For each depth class, rainfall duration is extracted using the empirical cumulative probability distribution (CDF) derived from the rainfall events within that class. According to a given rainfall depth and duration, the rainfall pattern is generated based on developed conditional CDF. Therefore, the final output of the RPG model is a rainfall pattern. In each model run, the normalized rainfall patterns are generated to provide rainfall patterns for the simulated rainfall events. A full description of the model and its operation are available in the study by Sharafati and Zahabiyoun [
Among various rainfall-runoff simulation models, HEC-HMS has evidenced its capacity to simulate flood patterns effectively [
The length of data (number of events) has a direct impact on calibrated values of HEC-HMS parameters. To quantify this source of uncertainty, the HEC-HMS parameters are considered as random variables. The uncertainty related to peak discharge and flood volume due to variability of HEC-HMS parameters and rainfall features are evaluated considering six different cases. A summary of random variables considered in different cases is presented in Table
Description of the cases considered in the present study.
Case | Base flow parameters | Loss parameters | Unit hydrograph parameter | Rainfall depth | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Case 1 | ||||||||||
Case 2 | ||||||||||
Case 3 | ||||||||||
Case 4 | ||||||||||
Case 5 | ||||||||||
Case 6 |
Case 1 considers only the base flow parameters (e.g., IF, RF, and RC) as random variables. Case 2 and Case 3 consider the loss parameters (e.g., IL, MD, WS, HC, and IA) and unit hydrograph (LT) as the stochastic parameters, respectively. Case 4 is defined to assess the uncertainty associated with rainfall depth. The randomness in characteristics of all the HEC-HMS parameters is considered in Case 5, but the variability of other rainfall features is ignored. Case 6 is used to assess the uncertainty associated with flood features due to the variability of rainfall depth and all the HEC-HMS parameters. The impact of rainfall duration on flood hydrograph is investigated for three different durations (12 hr, 18 hr, and 24 hr) in all the cases.
The PDFs of the HEC-HMS parameters are estimated from eight previously calibrated events [
To quantify the impact of uncertainty associated with model parameters and rainfall on flood hydrograph features (peak and volume), a new index named relative coefficient of variation (
To detect the most influencing hydrological parameters on flood hydrograph features, the average relative RCV (ARRCV) index for the
The four observed flood events (Figure
Observed hydrographs used to validate the model generated hydrographs.
The validity of flood hydrograph generation methodology is investigated by comparing observed flood volume and peak discharge within the significant band (90% confidence interval) of generated hydrographs (Figure
Comparison of observed and generated peak discharges and flood volumes.
The influence of uncertainty of base flow parameters on simulated flood hydrograph is investigated in case 1. All the hydrological parameters and rainfall depth are considered deterministic in this case. The RCV values of the peak discharge and flood volume for stochastic base flow parameters (e.g., IF, RF, and RC) for different rainfall durations (12 hr, 18 hr, and 24 hr) are presented in Figures
The impact of uncertainty in (a) base flow parameters on peak discharge, (b) base flow parameters on flood volume, (c) loss parameters on peak discharge, (d) loss parameters on flood volume, (e) unit hydrograph parameters on peak discharge, (f) unit hydrograph parameters on flood volume, (g) rainfall depth on peak discharge, and (h) rainfall depth on flood volume for different rainfall durations.
The RCV values of peak discharge and flood volume estimated to assess the impact of loss parameters on flood features are presented in Figures
Figures
In Case 4, all of the hydrological parameters are considered random, and therefore, it is used to compare the influence of the hydrological parameters on flood hydrograph features. Total of 10000 random hydrographs are generated based on the stochastic hydrological parameters for this purpose. Similar to cases 1–3, hydrographs are generated for different rainfall durations. Obtained results are compared with those obtained in cases 1–3 using ARRCV index.
Figure
The ARRCV (in percentage) obtained for different hydrological components (baseflow, loss, unit hydrograph, and rainfall depth).
To assess the impact of the hydrological parameters as well as rainfall depth on flood hydrograph features, those are generated randomly in case 5 and compared with the results obtained in case 4 (Figure
Uncertainties in flood hydrograph features, namely peak discharge and flood volume due to uncertainty in rainfall and hydrological model parameters are assessed in this study. The flood hydrograph is generated using the HEC-HMS model from rainfall events generated using RPG. The PDF is used to estimate the uncertainty in HEC-HMS parameters, including the loss, base flow, and unit hydrograph parameters. The RCV and ARRCV indices are used to estimate the influence of different parameters on flood peak and volume. The highest RCV values for flood peak and flood volume are found as 116 and 153, respectively, for a 12 hr duration rainfall event. This indicates that flood volume is more sensitive to uncertainty in the hydrological model and rainfall parameters compared to flood peak. Rainfall depth is found to have more influence on flood peak and volume compared to rainfall duration. The ARRCV index calculated for different hydrological parameters also indicates the highest impact of rainfall depth on uncertainty in flood hydrograph features.
As limitation of the current research, the proposed approach is examined on a single basin. However, employing more basins can provide a better perspective on the uncertainty analysis of flood features. Thus, this issue can be studied in future studies. Furthermore, several uncertainties sources related to the model configuration, data, and study period are missed in the current study, which can be considered to provide robust findings in future investigations.
The data used to support the findings of this study are available from the corresponding author upon request.
The authors declare have no conflicts of interest.