Physically-based Distributed Models Research Articles

Difference of rainfall-runoff models and effect on flood forecasting: A brief review

Selecting a rainfall-runoff model for use in flood forecasting is not a direct decision and actually may contain the selection of more than one. There are a range of rainfall-runoff models for flow forecasting. They range in type from transfer function (empirical black box), through lumped conceptual to more physically-based distributed models. The rainfall-runoff models also are often accompanied by updating techniques for taking account of recent measurements of flow so as to improve the accuracy of model predictions in real-time. Against this variety of available modelling techniques, this study improved understanding of the most important and well known rainfall-runoff models for flood forecasting and highlighting their similarities and differences. Six models are selected in this study: the Probability Distributed Moisture (PDM) model, the Isolated Event Model (IEM), the US National Weather Service Sacramento model, the Grid Model, the Transfer Function (TF) model and the Physically Realisable Transfer Function (PRTF) model. The first three are conceptual soil moisture accounting models, with the Grid Model having a distributed formulation, whilst the TF and PRTF are “black box” time-series models. Also new model for the forecasting (e.g neural network (NN), fuzzy rule-based are reviewed. An important feature of the use of rainfall-runoff models in a real-time forecasting environment is the ability to integrate recent observations of flow in order to develop forecast performance. The available methods for forecast updating are reviewed with specific reference to state correction and error prediction techniques.

Landslide Susceptibility Assessment Using the Scoops3D Model in a Tropical Mountainous Terrain

Many physically-based distributed models study the landslide occurrence using an infinite slope stability analysis, simulating a planar failure, which is not usually applicable to rotational failures and deep landslides. Recently, some three-dimensional distributed physically-based models have been developed that have been applied in different parts of the world. In this research, the Scoops3D model is implemented for a landslide susceptibility analysis in a tropical mountainous terrain of the Colombian Andes (Medellín, Colombia). In addition to identifying the areas susceptible to the occurrence of rotational landslides, the results of the safety factor are analyzed with the areas of associated critical failure surfaces to provide an interpretation and explanation of the simulation results. This is to have a better understanding of how the model works and to facilitate its implementation in landslide hazard assessment. The Scoops3D physicallybased model can be a very useful tool for mass movement risk management projects.

On the Operational Flood Forecasting Practices Using Low-Quality Data Input of a Distributed Hydrological Model

Low-quality input data (such as sparse rainfall gauges, low spatial resolution soil type and land use maps) have limited the application of physically-based distributed hydrological models in operational practices in many data-sparse regions. It is necessary to quantify the uncertainty in the deterministic forecast results of distributed models. In this paper, the TOPographic Kinematic Approximation and Integration (TOPKAPI) distributed model was used for deterministic forecasts with low-quality input data, and then the Hydrologic Uncertainty Processor (HUP) was used to provide the probabilistic forecast results for operational practices. Results showed that the deterministic forecasts by TOPKAPI performed poorly in some flood seasons, such as the years 1997, 2001 and 2008, despite which the overall accuracy of the whole study period 1996–2008 could be acceptable and generally reproduced the hydrological behaviors of the catchment (Lushi basin, China). The HUP model can not only provide probabilistic forecasts (e.g., 90% predictive uncertainty bounds), but also provides deterministic forecasts in terms of 50% percentiles. The 50% percentiles obviously improved the forecast accuracy of selected flood events at the leading time of one hour. Besides, the HUP performance decayed with the leading time increasing (6, 12 h). This work revealed that deterministic model outputs had large uncertainties in flood forecasts, and the HUP model may provide an alternative for operational flood forecasting practices in those areas with low-quality data.

Physically-based landslide susceptibility analysis using Monte Carlo simulation in a tropical mountain basin

ABSTRACT Physically-based distributed models are implemented for landslide susceptibility and hazard assessment around the world. Probabilistic methodologies are considered appropriate to study and quantify the uncertainties derived from the input parameters of these models. In this paper, three sets of Monte Carlo simulations, each one with 10,000 iterations, were applied for a slope stability analysis in a small basin of Envigado (Colombia), using the TRIGRS model, to characterise the uncertainty in the landslide assessment. Different parameters to determine the minimum number of realizations required to ensure a small variation in the failure probability were proposed and analyzed. The quality of the landslide susceptibility assessment was studied. Unexpected and probably erroneous results that may be common in the maps generated using this and other similar methodologies were identified and explained. Additionally, the distribution of the factor of safety was calculated for different grid cells of the basin, showing that the probability density function with the best adjustment to the frequency histogram of the factor of safety can vary between grid cells. The assumption of a normal distribution for the factor of safety would be inappropriate and would lead to miscalculations in this case study.

A conceptual data model coupling with physically-based distributed hydrological models based on catchment discretization schemas

Summary In hydrology, the data types, spatio-temporal scales and formats for physically-based distributed hydrological models and the distributed data or parameters may be different before significant data pre-processing or may change during hydrological simulation run time. A data model is devoted to these problems for sophisticated numerical hydrological modeling procedures. In this paper, we propose a conceptual data model to interpret the comprehensive, universal and complex water environmental entities. We also present an innovative integration methodology to couple the data model with physically-based distributed hydrological models (DHMs) based on catchment discretization schemas. The data model provides a reasonable framework for researchers of organizing and pre-processing water environmental spatio-temporal datasets. It also facilitates seamless data flow fluid and dynamic by hydrological response units (HRUs) as the core between the object-oriented databases and physically-based distributed hydrological models.

Simple Linear Modeling Approach for Linking Hydrological Model Parameters to the Physical Features of a River Basin

Physically-based distributed hydrological models are drawing more attention in recent years. Generally they have some parameters to be adjusted for satisfactory simulation performance. However, the parameter calibration procedure requires abundant input data. Fortunately, parameters of physical-based models are generally related to specific basin characteristics. If these relationships are generalized with caution, they could be used to simplify the parameter calibration procedure and facilitate estimation of parameters in ungauged basins. In this study, comprehensive tests between sensitive hydrological model parameters and all available basin characteristics, including two new defined characteristics, are done to exploit their potential relationships using Cluster Analysis. Then linear regression model is adopted to fit a reliable relation for each sensitive parameter respectively. The matrix condition number is used to diagnose the multicollinearity in explanatory variables to ensure effective and robust relations. Moreover, two new characteristics are defined to make better use of characteristics consisting of component data. Their remarkable performance suggests their irreplaceable role in fitting a reliable and robust relationship. When the relations for sensitive parameters are fitted out, they are used to transfer parameters to independent sub-basin from other sub-basins. The leave-one-out cross validation is used to evaluate transferring performances. The BTOPMC model is used to apply the above approach in the Yalong River Basin, Southwest China. Thus the samples of this study are various and representative. It shows that the transferred parameters have satisfactory performance, which suggests that the proposed approach is effective with application prospect.

Modeller subjectivity and calibration impacts on hydrological model applications: An event-based comparison for a road-adjacent catchment in south-east Norway

Identifying a ‘best’ performing hydrologic model in a practical sense is difficult due to the potential influences of modeller subjectivity on, for example, calibration procedure and parameter selection. This is especially true for model applications at the event scale where the prevailing catchment conditions can have a strong impact on apparent model performance and suitability. In this study, two lumped models (CoupModel and HBV) and two physically-based distributed models (LISEM and MIKE SHE) were applied to a small catchment upstream of a road in south-eastern Norway. All models were calibrated to a single event representing typical winter conditions in the region and then applied to various other winter events to investigate the potential impact of calibration period and methodology on model performance. Peak flow and event-based hydrographs were simulated differently by all models leading to differences in apparent model performance under this application. In this case-study, the lumped models appeared to be better suited for hydrological events that differed from the calibration event (i.e., events when runoff was generated from rain on non-frozen soils rather than from rain and snowmelt on frozen soil) while the more physical-based approaches appeared better suited during snowmelt and frozen soil conditions more consistent with the event-specific calibration. This was due to the combination of variations in subsurface conditions over the eight events considered, the subsequent ability of the models to represent the impact of the conditions (particularly when subsurface conditions varied greatly from the calibration event), and the different approaches adopted to calibrate the models. These results indicate that hydrologic models may not only need to be selected on a case-by-case basis but also have their performance evaluated on an application-by-application basis since how a model is applied can be equally important as inherent model structure.

Análise de sensibilidade dos parâmetros do modelo SWAT e simulação dos processos hidrossedimentológicos em uma bacia no agreste nordestino

O processo erosivo constitui-se na principal causa de degradação dos solos e é acelerado por intervenções humanas nas bacias hidrográficas, resultando em prejuízos ao setor agrícola e ao meio ambiente. Para a previsão dos impactos de ações antrópicas ou de mudanças climáticas sobre os processos hidrossedimentológicos, os modelos distribuídos e de base física têm sido bastante eficazes. Neste estudo, aplicou-se o modelo SWAT à sub-bacia do Rio Japaratuba Mirim, em Sergipe, nas seções de medição das estações Fazenda Pão de Açúcar (PA) e Fazenda Cajueiro (CJ), com áreas de contribuição respectivamente de 137,3 e 277,8 km², visando à parametrização das duas bacias aninhadas por meio dos processos de calibração e de validação. Foram realizadas simulações do escoamento na bacia menor (PA), a partir dos parâmetros calibrados na bacia maior (CJ), e, reciprocamente, do escoamento na bacia maior (CJ). Em seguida, foram também realizadas simulações do escoamento e da produção de sedimentos para o período de dados de vazão disponíveis (1985 a 2000), para a seção CJ. Os resultados evidenciaram que o modelo calibrado simulou bem o escoamento superficial e conseguiu prever, de forma coerente, a produção de sedimentos com base nos 12 parâmetros mais sensíveis do modelo. A transposição dos parâmetros da bacia maior para a menor e, vice-versa, resultou em índices de Eficiência de Nash-Sutcliffe (NSE) e de Tendência Percentual (PBIAS), considerados satisfatórios para o primeiro caso e insatisfatório para o segundo.

Is there a need for hydrological modelling in decision support systems for nuclear emergencies.

This paper discusses the role of hydrological modelling in decision support systems for nuclear emergencies. In particular, most recent developments such as, the radionuclide transport models integrated in to the decision support system RODOS will be explored. Recent progress in the implementation of physically-based distributed hydrological models for operational forecasting in national and supranational centres, may support a closer cooperation between national hydrological services and therefore, strengthen the use of hydrological and radiological models implemented in decision support systems.

Increasing vulnerability to extreme floods and societal needs of hydrological forecasting

Recent climatic variability seems to be intensifying the hydrological extremes resulting in devastating hydrological hazards, especially floods all over the world, including the arid regions. Disasters are always more tragic in poor countries than in rich ones, as more poor people seem to suffer and die, and the economic losses have a greater impact. On the other hand, in advanced countries, although flood control can reduce the number of deaths, it cannot always reduce economic losses despite the large public investments that have long been applied. This is due to the ever increasing damage potential in a basin where reliance on safety has been created by flood control measures. It is proven that proper prediction and preparedness are the only practical means that can be relied upon to reduce the death toll and economic losses. In a time of climatic variation, physically-based distributed models are to be pursued to obtain reasonable predictions, since the climate is non-stationary and both basin land use and water use are constantly changing. The proper integration of a distributed watershed model with atmospheric circulation models, four-dimensional data assimilation and mesoscale hydrometeorological nested models is the direction for hydrologists to be heading.

The influence of different methods to derive soil hydraulic properties on the uncertainty of various model outputs of a distributed hydrological model

Measuring soil hydraulic properties is a costly and time consuming task, especially when these are needed for large areas in e.g. application of physically-based distributed hydrological models on a catchment scale. The use of pedo-transfer-functions (PTF) to derive the soil hydraulic parameters can be a suitable solution. In this study the spatial uncertainty involved with the soil hydraulic parameters, (i) derived from lab measurements, (ii) calculated with PTF using USDA texture figures. and (iii) calculated with PTF using USDA texture figures, is compared and evaluated on their resulting uncertainty on the different model outputs of a physically-based distributed hydrological model. A Latin Hypercube Simulation with MIKE SHE as the distributed hydrological model was performed for a small catchment in Germany using the soil hydraulic parameters from the three methods. Depending on the fact whether a certain model output is sensitive to the soil hydraulic parameters, different results can be found for the simulated discharge, groundwater head and soil water content.

A study on subcatchment scale for a distributed runoff model

Physically-based distributed runoff models are more advantageous than concentrated conceptual models for predicting the hydrologic response of ungauged basins and the changing response resulting from changes in land-use. In constructing distributed models, the river basin must be divided into reasonable smaller subcatchments corresponding to a characteristic scale for rainfall-runoff response. Thus, one of the still-unsolved problems in distributed models is the estimation of the appropriate subcatchment scale. This paper focuses only on the appropriate subcatchment scale considering the channel network geomorphology, although the effects of subcatchment scale on the overall river basin response are important. For this purpose, some new stochastic functions are proposed for discussing the hydrologic response of channel networks. By applying these functions into a simplified distributed runoff model, the relationship between the hydrologic response of the river basin and variation of the channel network geomorphology is shown.

Sensitivity analysis of the Systeme Hydrologique Europeen for an upland catchment

In order to study the accuracy with which parameter sets for physically-based distributed catchment models must be prepared and calibrated, a sensitivity analysis of the Système Hydrologique Européen was carried out based on simulations of two streamflow hydrographs for an upland catchment in mid-Wales. A single parameter was varied at a time and for each hydrograph the simulation sensitivity was assessed quantitatively in terms of the changes in peak discharge and in the root mean square value of the differences between measured and simulated discharges taken at intervals through the hydrograph. An important qualitative assessment was provided through interpretation of the changes based on physical reasoning. The results show that the simulations can be as sensitive to model grid spacing and time step as to catchment parameters and that these “structural” parameters should therefore be small by comparison with the scales of the variations which they are used to represent. Those catchment parameters to which the simulations are most sensitive (soil and flow resistance coefficients) can be evaluated with sufficient accuracy by point measurements at a few representative field sites, while the least important parameters (vegetation coefficients) can be evaluated using data from the literature. The scope for achieving equally satisfactory calibrations based on different combinations of parameter values is limited as long as several different hydrographs are considered. Similarly, for this particular application, the spatial distributions in rainfall and soil parameters have a relatively minor effect on the simulations.