Sarka Blazkova

Flood frequency estimation by continuous simulation for a gauged upland catchment (with uncertainty)

This paper explores the possibility of deriving frequency distributions of extreme discharges by continuous simulation. The rainfall-runoff model TOPMODEL is applied within the Generalised Likelihood Uncertainty Estimation (GLUE) framework to the River Wye catchment, Plynlimon, Wales, using a 21-year period of rainfall and discharge observations. Rejection of non-behavioural parameter sets is achieved through an evaluation of both annual maximum discharge and continuous hydrograph simulation. Annual maximum peak timings and rankings are also considered. It is demonstrated that, within the prescribed limits, TOPMODEL can adequately achieve both flood frequency and continuous simulation modelling goals. Extension of the flood frequency estimations beyond the upper limit of the observed series is attained through the coupling of behavioural TOPMODEL sets with those of a stochastic rainfall generator for 1000-year simulation periods using hourly time steps. The rainfall model is conditioned on the observed rainfall frequency statistics for different storm duration classes, also within the GLUE framework.

A limits of acceptability approach to model evaluation and uncertainty estimation in flood frequency estimation by continuous simulation : Skalka catchment, Czech Republic.

In this study continuous simulation flood frequency predictions on the Skalka catchment in the Czech Republic (672 km 2, range of altitudes from 460 to 1041 m above sea level), are compared against summary information of rainfall characteristics, the flow duration curve, and the frequency characteristics of flood discharges and snow water equivalent using the generalized likelihood uncertainty estimation limits of acceptability approach outlined by Beven (2006). Limits of acceptability have been defined, prior to running the Monte Carlo model realizations for subcatchment rainfalls, discharges (using rating data) at 5 sites within the catchment, and snow water equivalent in 13 snow zones, 4 of which have observed data. Flood frequency and flow duration data at the outlet of the whole catchment are not used in the evaluation but are used to test the predictions. In order to get sufficient behavioral models to assess adequately the prediction uncertainty it was necessary to refine the model structure, sample the model space more densely, and, in the end, relax the limits of acceptability to allow for a strong realization effect in predicted flood frequencies. We use a procedure of scoring deviations relative to the limits of acceptability to identify the minimum extension of the limits across all criteria to obtain a sample of 4192 parameter sets that were accepted as potentially useful in prediction. Results show that individual model realizations, with the same parameter values, of similar length to the observations can vary significantly in acceptability. Long-term simulations of 10,000 years for retained models were used to obtain uncertain estimates of the 1000 year peak and associated flood hydrographs required for the assessment of dam safety at the catchment outlet.

Flood frequency estimation by continuous simulation for a catchment treated as ungauged (with uncertainty)

A general methodology for flood frequency estimation based on continuous simulation is here applied to a gauged site in the Czech Republic treated as if it was ungauged. In this implementation, stochastic temperature and precipitation models are used to drive TOPMODEL to simulate stream discharges. The coupled model parameters are varied randomly across specified ranges using Monte Carlo simulation. The results from a sample of 48,600 simulations each of length 100 years using an hourly time step are conditioned on low return period regionalized flood frequency, snow water equivalent, and flow duration curve information. Performance measures for each predicted variable are combined using fuzzy inference and simulations considered as nonbehavioral are rejected. 10,000-year simulations are made with the remaining 2281 behavioral simulations to produce prediction limits for flood magnitudes and other response variables at different return periods. The results are checked against a historical series of annual maximum discharges available at the site for a period before it was destroyed by the construction of a dam. The results compare well and appear to give more realistic prediction bounds than statistical extrapolations based on the Wakeby distribution, particularly at longer return periods.

Testing the distributed water table predictions of TOPMODEL (allowing for uncertainty in model calibration): The death of TOPMODEL?

The distributed predictions of the original version of TOPMODEL are here compared with distributed observations of water table levels in the Uhlirska catchment in the Jizera Mountains, Czech Republic. The calibration of the model has been carried out within the GLUE framework, which allows the estimation of uncertainties in predicting the distributed patterns of the water table at different times. Many of the water table levels are predicted within the limits of uncertainty, but it is shown that the predictions could be improved by the calculation of a local effective transmissivity value (or local upslope contributing areas) at each observation site. These effective transmissivities show a similar relationship to the topographic index as found in a previous study of a small catchment in Norway. Some of the anomalies can be explained by deficiencies in the topographic analysis but this may also be an indication of possible structural deficiencies in the model. Interpretation is, however, difficult, and it remains to be seen whether these anomalies might be avoided in more dynamic distributed models.

Discharge‐dependent pollutant dispersion in rivers: Estimation of aggregated dead zone parameters with surrogate data

Much has been done to mitigate the effects of intermittent discharges of pollutants; however, pollution incidents still occur, and the downstream transport and dispersion of pollutants must be predicted. The application of transient storage models based on the advection dispersion equation is often limited by the strong dependence of the parameters on changes in discharge. In this paper a methodology is outlined for estimating the parameters of the simple aggregated dead zone model using surrogate data derived from continuous water quality measurements such as conductivity, including a full treatment of the errors and prediction uncertainties within a Bayesian framework. This methodology is demonstrated in the prediction of a tracer experiment on a reach of the river Elbe in the Czech Republic.

Uncertainty in flood estimation

The objective of this contribution is to form a clear picture of uncertainties we encounter in flood estimation, including both real-time flood forecasting and simulation for flood risk estimation. In simulation, we prefer the thesis of equifinality to obtain global optima. Many models producing acceptable simulations can be considered as multiple working hypotheses about the system process representations. Some of those hypotheses might later be confirmed or rejected, given additional data. In GLUE (Generalized Likelihood Uncertainty Estimation) the parameter sets are sampled randomly from physically reasonable ranges, often using uniform sampling where there is no strong information about prior expectations of parameter values. The parameter sets are then used to generate different realizations of the model outputs, which are then evaluated using some criteria (measures of likelihood) to provide a weight associated with each parameter set. Likelihood here is used in a much broader sense than in statistical inference. If some limits of effective observation error can be specified prior to running any simulations, models predicting outside of those limits can then be rejected as non-behavioural. Thus, any model evaluation of this type needs to take account of the multiple sources of model error more explicitly. This, however, is difficult for realistic cases. The procedure for the GLUE methodology is illustrated in examples. Usability for practical problems is suggested and future development is outlined.

Comparison of saturated areas mapping methods in the Jizera Mountains, Czech Republic

Abstract Understanding and modelling the processes of flood runoff generation is still a challenge in catchment hydrology. In particular, there are issues about how best to represent the effects of the antecedent state of saturation of a catchment on runoff formation and flood hydrographs. This paper reports on the experience of mapping saturated areas using measured water table by piezometers and more qualitative assessments of the state of the moisture at soil surface or immediately under it to provide information that can usefully condition model predictions. Vegetation patterns can also provide useful indicators of runoff source areas, but integrated over much longer periods of time. In this way, it might be more likely that models will get the right predictions for the right reasons.

Influence of afforestation on water regime in Jizera Catchments, Czech Republic

This paper studies the influence of afforestation on the water regime in two catchments in the Jizera Mountains that are similar in size and altitude but have different afforestation pattern. In this paper a range of different modelling tools is used to establish whether the differences in catchment water regime can be quantified and attributed to differences in catchment characteristics. Frequency analysis of low and high flows and a number and duration of flows over a threshold value are used to look for the differences in flow regime in both catchments. Low flow conditions are modelled using the Wittenberg nonlinear store approach. A rainfall-runoff process is modelled using a Data Based Mechanistic approach. The results indicate that the differences in the catchment response to external climatic factors outweigh the influence of land use apart from the low flows, where the changes in the response might be attributed to afforestation.

Analysis of the time series of waste water quality at the inflow of the wastewater treatment plant and transfer functions

Abstract Time series of the daily total precipitation, daily wastewater discharges and daily concentrations and pollution loads of BOD5, COD, SS, N-NH4, Ntot and Ptot were analyzed at the inflow to the wastewater treatment plant (WWTP) for a larger city in 2004-2009 (WWTP is loaded by pollution from 435,000 equivalent inhabitants). The time series of the outflow from a WWTP was also available for 2007. The time series of daily total precipitation, daily wastewater discharges, concentrations and pollution loads at the inflow and outflow from the WWTP were standardized year by year to exclude a long-term trend, and periodic components with a period of 7 days and 365 days (and potentially also 186.5 days) were excluded from the standardized series. However, these two operations eliminated only a small part of the variance; there was a substantial reduction in the variance only for ammonium nitrogen and total nitrogen at the inflow and outflow from a WWTP. The relationship between the inflow into a WWTP and the outflow from a WWTP for the concentrations and pollution loads was described by simple transfer functions (SISO models) and more complicated transfer functions (MISO models). A simple transfer function (SISO model) was employed to describe the relationship between the daily total precipitation and the wastewater discharge.

Transport and Dispersion in Large Rivers: Application of the Aggregated Dead Zone Model

The Aggregated Dead Zone (ADZ) model developed by Peter Young and Tom Beer is here applied to the analysis of tracer data from larger rivers. The model provides excellent fits to the observed concentrations, with a dispersive fraction parameter that varies relatively little with discharge. It is also shown how the information on transport and dispersion at different discharges can be augmented by pollution incident and continuously logged water quality data. The model can then be applied to predict the downstream dispersion of pollutants at any arbitrary discharge, taking account of the uncertainty in estimating the ADZ model parameters. Further work remains to be done on relating the parameters of the model to the physical and hydraulic characteristics of river reaches and in gathering data on the gain factor for different pollutants.