No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
The effects of spatial discretization and model parameterization on the prediction of extreme runoff characteristics
Abstract
Water resources management in mesoscale river basins requires, among other things, reliable predictions on extreme runoff characteristics such as magnitude and frequency of floods and droughts. Hydrologic models are increasingly used for these prediction purposes. Outputs of these models, however, are sensitive to various factors like the spatial representation of hydrologic processes, the parameterization method, and the type of estimator used for calibration. This study aimed to investigate the possible effects of these factors on extreme runoff characteristics derived from simulated streamflow. For this purpose, lumped and distributed versions of the conceptual mesoscale hydrologic model (mHM) were implemented in 22 German basins ranging in size from 58 to 4000 km2. The distributed mHM version was, in turn, parameterized with hydrological response units (HRU) and multiscale parameter regionalization (MPR) methods. Free parameters of both model versions were calibrated with three objective functions emphasizing high flows, low flows, and a combination of both. Six extreme runoff characteristics were derived from daily streamflow simulations for winter and summer. Results indicated that the model performance evaluated with both daily streamflow and seasonal runoff characteristics was sensitive to the type of estimator, the spatial discretization, and the parameterization method employed. The lumped version exhibited the highest sensitivity to previous factors and the least performance, whereas the opposite behavior was noticed for the distributed version parameterized with the MPR technique. Furthermore, the efficiency of the model parameterized with MPR were higher than that obtained with the HRU parameterization, in particular, when the model was evaluated in locations not used for calibration.
... This is sometimes referred to as semi-distributed approach, often being the preferred choice in practical applications (e.g. Euser et al., 2015;Kumar et al., 2010). Although no clear definition exists, this family of discretisation typically consists of a hierarchical multi-scale scheme dividing a hydrological basin into several subbasins which in turn contain irregularly shaped computational elements being hydrologically uniform entities (e.g. ...
... The role of detail of discretisation (i.e. the spatial resolution) of hydrological models, partly also as a user decision during the pre-processing process, has long been acknowledged in numerous studies (e.g. Euser et al., 2015;González et al., 2016;Haghnegahdar et al., 2015;Han et al., 2014;Kumar et al., 2010;Wood et al., 1988). For grid-based models, this influence has been thoroughly assessed (e.g. ...
... Thus, several studies examined the impact of discretisation complexity on model performance. They showed that (semi-)distributed models are usually more suitable for the representation of landscape heterogeneity and the exploration of hydrological processes, and are also more able to reproduce observed discharge dynamics than lumped models (Euser et al., 2015;Kumar et al., 2010). However, there is a threshold of subdivision level above which no more improvements can be achieved (Haghnegahdar et al., 2015;Han et al., 2014;Wood et al., 1988). ...
Hydrological models are important tools for the simulation and quantification of the water cycle. They therefore aid in the understanding of hydrological processes, prediction of river discharge, assessment of the impacts of land use and climate changes, or the management of water resources. However, uncertainties associated with hydrological modelling are still large. While significant research has been done on the quantification and reduction of uncertainties, there are still fields which have gained little attention so far, such as model structural uncertainties that are related to the process implementations in the models. This holds especially true for complex process-based models in contrast to simpler conceptual models. Consequently, the aim of this thesis is to improve the understanding of structural uncertainties with focus on process-based hydrological modelling, including methods for their quantification. To identify common deficits of frequently used hydrological models and develop further strategies on how to reduce them, a survey among modellers was conducted. It was found that there is a certain degree of subjectivity in the perception of modellers, for instance with respect to the distinction of hydrological models into conceptual groups. It was further found that there are ambiguities on how to apply a certain hydrological model, for instance how many parameters should be calibrated, together with a large diversity of opinion regarding the deficits of models. Nevertheless, evapotranspiration processes are often represented in a more physically based manner, while processes of groundwater and soil water movement are often simplified, which many survey participants saw as a drawback. A large flexibility, for instance with respect to different alternative process implementations or a small number of parameters that needs to be calibrated, was generally seen as strength of a model. Flexible and efficient software, which is straightforward to apply, has been increasingly acknowledged by the hydrological community. This work further elaborated on this topic in a twofold way. First, a software package for semi-automated landscape discretisation has been developed, which serves as a tool for model initialisation. This was complemented by a sensitivity analysis of important and commonly used discretisation parameters, of which the size of hydrological sub-catchments as well as the size and number of hydrologically uniform computational units appeared to be more influential than information considered for the characterisation of hillslope profiles. Second, a process-based hydrological model has been implemented into a flexible simulation environment with several alternative process representations and a number of numerical solvers. It turned out that, even though computation times were still long, enhanced computational capabilities nowadays in combination with innovative methods for statistical analysis allow for the exploration of structural uncertainties of even complex process-based models, which up to now was often neglected by the modelling community. In a further study it could be shown that process-based models may even be employed as tools for seasonal operational forecasting. In contrast to statistical models, which are faster to initialise and to apply, process-based models produce more information in addition to the target variable, even at finer spatial and temporal scales, and provide more insights into process behaviour and catchment functioning. However, the process-based model was much more dependent on reliable rainfall forecasts. It seems unlikely that there exists a single best formulation for hydrological processes, even for a specific catchment. This supports the use of flexible model environments with alternative process representations instead of a single model structure. However, correlation and compensation effects between process formulations, their parametrisation, and other aspects such as numerical solver and model resolution, may lead to surprising results and potentially misleading conclusions. In future studies, such effects should be more explicitly addressed and quantified. Moreover, model functioning appeared to be highly dependent on the meteorological conditions and rainfall input generally was the most important source of uncertainty. It is still unclear, how this could be addressed, especially in the light of the aforementioned correlations. The use of innovative data products, e.g.\ remote sensing data in combination with station measurements, and efficient processing methods for the improvement of rainfall input and explicit consideration of associated uncertainties is advisable to bring more insights and make hydrological simulations and predictions more reliable.
... The mesoscale Hydrologic Model (mHM; Kumar et al., 2010;Samaniego et al., 2010b) accounts for diverse processes of 5 the hydrological cycle: Canopy interception, evapotranspiration, snow, soil moisture dynamics, overland flow, infiltration, interflow, subsurface storage, groundwater recharge, baseflow, discharge attenuation as well as flood routing. The mHM is conceptualised on the basis of grid cells, and has been applied to a wide range of mesoscale river catchments (10 1 − 10 4 km 2 ; Kumar et al., 2010;Samaniego et al., 2010aSamaniego et al., , b, 2011Cuntz et al., 2015;Rakovec et al., 2016a, b). ...
... The mesoscale Hydrologic Model (mHM; Kumar et al., 2010;Samaniego et al., 2010b) accounts for diverse processes of 5 the hydrological cycle: Canopy interception, evapotranspiration, snow, soil moisture dynamics, overland flow, infiltration, interflow, subsurface storage, groundwater recharge, baseflow, discharge attenuation as well as flood routing. The mHM is conceptualised on the basis of grid cells, and has been applied to a wide range of mesoscale river catchments (10 1 − 10 4 km 2 ; Kumar et al., 2010;Samaniego et al., 2010aSamaniego et al., , b, 2011Cuntz et al., 2015;Rakovec et al., 2016a, b). Gridded information is implemented in mHM at three levels: morphology (level 0), hydrology (level 1), meteorology (level 2), with l 0 l 1 ≤ l 2 10 denoting the relative sizes of the grid cells at the respective data level (Kumar et al., 2010). ...
... The mHM is conceptualised on the basis of grid cells, and has been applied to a wide range of mesoscale river catchments (10 1 − 10 4 km 2 ; Kumar et al., 2010;Samaniego et al., 2010aSamaniego et al., , b, 2011Cuntz et al., 2015;Rakovec et al., 2016a, b). Gridded information is implemented in mHM at three levels: morphology (level 0), hydrology (level 1), meteorology (level 2), with l 0 l 1 ≤ l 2 10 denoting the relative sizes of the grid cells at the respective data level (Kumar et al., 2010). ...
Diagnostics of hydrological models is pivotal for a better understanding of catchment functioning. The analysis of dominating parameters for the simulation of streamflow plays a key role for region specific model diagnostics, model calibration or parameter transfer. A major challenge in this analysis of parameter sensitivity is the assessment of both temporal and spatial differences of parameter influences on simulated streamflow response. A methodical approach is presented, wherein a two-tiered global sensitivity analysis on a spatially distributed hydrological model is applied to 14 mesoscale headwater catchments of the river Ruhr in western Germany. The analysis of parameter sensitivity is geared towards two complementary forms of streamflow response targets. The analysis of the temporal dynamics of parameter sensitivity (TEDPAS) is contrasted with sensitivity analysis directed to hydrological fingerprints, i.e. temporally independent and temporally aggregated characteristics of streamflow (INDPAS). The two-tiered approach allows to discern a clarified sensitivity pattern pinpointed to diverse response characteristics, to detect regional differences and to reveal the regional relevance of the response target. Small local differences in the hydroclimatic and topographic setting of the headwaters lead to slight differences in the hydrological functioning, which was revealed by gradual differences in TEDPAS and INDPAS.
... The numerical experiments are conducted in the upper Neckar catchment (Fig. 2) that was extensively investigated in previous hydrological studies (Kumar et al., 2010;Samaniego et al., 2010a, b;Wöhling et al., 2013b). This catchment is located in the central uplands of Germany and comprises a catchment area of approximately 4000 km 2 . ...
... All these data are discretized to a spatial resolu-tion of 100 m × 100 m. Readers interested in more details on data set and the processing may refer to Kumar et al. (2010), Samaniego et al. (2010b) and Zink et al. (2017). The spatial distributions of cumulative rain, potential evapotranspiration, land use and the mean annual leaf area index are shown in the Supplement (see Fig. S1). ...
... The model was calibrated and validated in previous studies showing very good capability to match streamflow measurements at catchment of different sizes (Kumar et al., 2010(Kumar et al., , 2013Samaniego et al., 2010b;Wöhling et al., 2013b). The same parameterization is used for the present study. ...
Soil properties show high heterogeneity at different spatial scales and their correct characterization remains a crucial challenge over large areas. The aim of the study is to quantify the impact of different types of uncertainties that arise from the unresolved soil spatial variability on simulated hydrological states and fluxes. Three perturbation methods are presented for the characterization of uncertainties in soil properties. The methods are applied on the soil map of the upper Neckar catchment (Germany), as an example. The uncertainties are propagated through the distributed mesoscale hydrological model (mHM) to assess the impact on the simulated states and fluxes. The model outputs are analysed by aggregating the results at different spatial and temporal scales. These results show that the impact of the different uncertainties introduced in the original soil map is equivalent when the simulated model outputs are analysed at the model grid resolution (i.e. 500 m). However, several differences are identified by aggregating states and fluxes at different spatial scales (by subcatchments of different sizes or coarsening the grid resolution). Streamflow is only sensitive to the perturbation of long spatial structures while distributed states and fluxes (e.g. soil moisture and groundwater recharge) are only sensitive to the local noise introduced to the original soil properties. A clear identification of the temporal and spatial scale for which finer-resolution soil information is (or is not) relevant is unlikely to be universal. However, the comparison of the impacts on the different hydrological components can be used to prioritize the model improvements in specific applications, either by collecting new measurements or by calibration and data assimilation approaches. In conclusion, the study underlines the importance of a correct characterization of uncertainty in soil properties. With that, soil maps with additional information regarding the unresolved soil spatial variability would provide strong support to hydrological modelling applications.
... Currently, mHM has been evaluated in more than one hundred basins in Germany ranging from 4 km 2 to 47 000 km 2 Kumar et al., 2010Kumar et al., , 2013b. This model is driven by disaggregated fields of daily forcings such as precipitation, temperature, and potential evapotranspiration. ...
... data. For a detailed description on data processing and setting up mHM in several river basins, interested readers may refer to Samaniego et al. (2010); Kumar et al. (2010). Previous data sets were re-sampled on a common spatial resolution of 100 ⇥ 100 m denoted as level-0. ...
... First, in every major river basin depicted in Figure 3.1, the dynamically dimensioned search algorithm (Tolson and Shoemaker, 2007) was employed to find good sets of global parameters which exhibit an acceptable model e ciency [e.g. Nash-Sutclife-E ciency of at least 0.75] during the evaluation period (for details refer to Kumar et al., 2010Kumar et al., , 2013b. In the next step, all parameter sets found for a given basin were transferred to the remaining ones. ...
Droughts are worldwide the second most severe natural disaster beside floods. In Europe, droughts are the costliest natural disasters with average expenses of 621 million EUR per event. The last severe drought event took place in 2003. It induced an agro-economic loss of 1.5 billion EUR in Germany alone. Such economical losses emphasize the need of an operational system for monitoring agricultural droughts in order to mitigate their negative consequences.
Observation-based monitoring of agricultural droughts, which are characterized by soil moisture deficits, is technically and economically not feasible on regional to national scales. Hydrologic modeling is the prime alternative to estimate soil moisture availability on large spatial domains. Such models are driven by meteorological observations and predict hydrological fluxes and states, such as soil moisture or evapotranspiration. Predictions of hydrologic models underlie several sources of uncertainties. These uncertainties arise from input data, model structure, initial conditions, and model parameters. The implications of parametric uncertainty to hydrologic predictions are analyzed herein.
The main objective of this work is to develop a monitoring system for agricultural droughts in Germany. The development of such a system includes several challenges. First, a spatially continuous dataset of soil moisture for entire Germany is derived from modeling. The parametric uncertainty of such hydrologic predictions is taken into account. Second, the propagation of parametric uncertainty of soil moisture to the identification of drought characteristics is estimated in order to evaluate the uncertainty inherent to such a monitoring system. Third, an approach to reduce the parametric uncertainty by using satellite retrieved land surface temperature data is investigated. And forth, an operational system providing drought information in near-real time is developed and implemented.
... The numerical experiments are conducted in the upper Neckar catchment ( Figure 2) that was extensively investigated in 30 previous hydrological studies (Kumar et al., 2010;Samaniego et al., 2010aSamaniego et al., , 2010bWöhling et al., 2013b). This catchment is located in the central uplands of Germany and comprises a catchment area of approximately 4000 km 2 . ...
... All these data are discretized to a spatial resolution of 100 x 100 m 2 . Readers interested in more details on data-set and the processing may refer to Kumar et al. (2010), Samaniego et al. (2010b) and Zink et al. (2016). The spatial distributions of cumulative rain, potential evapotranspiration, land use and the mean annual leaf area index are shown in the 15 supplementary material (see Figure S1). ...
... The model was calibrated and validated in previous studies showing very good capability to match streamflow measurements at catchment of different sizes (Kumar et al., 2010(Kumar et al., , 2013Samaniego et al., 2010b;Wöhling et al., 2013b). ...
Soil properties show high heterogeneity at different spatial scales and their correct characterization remains a crucial challenge over large areas. The aim of the study is to quantify the impact of different types of uncertainties that arise from the unresolved soil spatial variability on simulated hydrological states and fluxes. Three perturbation methods are presented for the characterization of the uncertainties in soil properties. The methods are applied at the soil map of the upper Neckar catchment (Germany), as example. The uncertainties are propagated based on the distributed hydrological model mHM to assess the impact of the simulated state and fluxes. The model outputs are analysed by aggregating the results at different spatial and temporal scales. These results show that the impact of the different uncertainties introduced in the original soil map is equivalent when the simulated model outputs are analysed at the model grid resolution (i.e., 500 m). However, several differences are identified by aggregating state and fluxes at different spatial scales (by subcatchments of different sizes or coarsening the grid resolution). Streamflow is only sensitive to the perturbation of long spatial structures while distributed state and fluxes (e.g., soil moisture and groundwater recharge) are only sensitive to the local noise introduced to the original soil properties. A clear identification of the temporal and spatial scale for which finer resolution soil information is (or not) relevant is unlikely to be universal. However, the comparison of the impacts on the different hydrological components can be used to prioritize the model improvements in specific applications, either by collecting new measurements or by calibration and data assimilation approaches. In conclusion, the study underlines the importance of a correct characterization of the uncertainty in soil properties. With that, soil map with additional information regarding the unresolved soil spatial variability would provide a strong support to hydrological modelling applications.
... Traditional parameter estimation of conceptual models relies on the availability of calibration data, which, however, are frequently not available for the time period or the spatio-temporal resolution of interest. A wide range of regionalization techniques for model parameters and hydrological signatures have been developed to avoid calibration in such data scarce environments (e.g., Bárdossy, 2007;Yadav et al., 2007;Perrin et al., 2008;Zhang et al., 2008;Kling and Gupta, 2009;Samaniego et al., 2010;Kumar et al., 2010;Wagener and Montanari, 2011;Kapangaziwiri et al., 2012;Viglione et al., 2013). However, it is challenging to identify suitable functional relationships between catchment characteristics and model parameters (e.g., Merz and Blöschl, 2004;Kling and Gupta, 2009), and only recently did Kumar et al. (2010Kumar et al. ( , 2013b show that multi-scale parameter regionalization (MPR) can yield global parameters that perform consistently over different catchment scales. ...
... A wide range of regionalization techniques for model parameters and hydrological signatures have been developed to avoid calibration in such data scarce environments (e.g., Bárdossy, 2007;Yadav et al., 2007;Perrin et al., 2008;Zhang et al., 2008;Kling and Gupta, 2009;Samaniego et al., 2010;Kumar et al., 2010;Wagener and Montanari, 2011;Kapangaziwiri et al., 2012;Viglione et al., 2013). However, it is challenging to identify suitable functional relationships between catchment characteristics and model parameters (e.g., Merz and Blöschl, 2004;Kling and Gupta, 2009), and only recently did Kumar et al. (2010Kumar et al. ( , 2013b show that multi-scale parameter regionalization (MPR) can yield global parameters that perform consistently over different catchment scales. In a further study they successfully transferred parameters obtained by the MPR technique to ungauged catchments in Germany and the USA (Kumar et al., 2013a). ...
... HRUs are units within a catchment, characterized by a different hydrological function and can be represented by different model structures or parameters. In most cases HRUs are defined based on soil types, land cover and other physical catchment characteristics (e.g., Knudsen et al., 1986;Flügel, 1995;Grayson and Blöschl, 2000;Krcho, 2001;Winter, 2001;Scherrer and Naef, 2003;Uhlenbrook et al., 2004;Wolock et al., 2004;Pomeroy et al., 2007;Scherrer et al., 2007;Schmocker-Fackel et al., 2007;Efstratiadis and Koutsoyiannis, 2008;Lindström et al., 2010;Nalbantis et al., 2011;Kumar et al., 2010). ...
... Estimating each of the 28 mHM parameters for each grid modeling cell through calibration will result in over-parameterization [69]. To reduce the number of free calibrated parameters vis à vis the prediction uncertainty, MPR was employed to translate high-resolution input data variables into model parameters using transfer functions and upscaling operators [70]. ...
... In the final stage, these high-resolution parameters are upscaled to produce fields of effective parameters at the required hydrologic modeling spatial scale (Level-1) using upscaling operators such as arithmetic mean, geometric mean, or harmonic mean. Kumar et al. [69] summarized these two steps as follows: ...
Hydrologic modeling in Nigeria is plagued by non-existent or paucity of hydro-metrological/morphological records, which has detrimental impacts on sustainable water resource management and agricultural production. Nowadays, freely accessible remotely sensed products are used as inputs in hydrologic modeling, especially in regions with deficient observed records. Therefore, it is appropriate to utilize the fine-resolution spatial coverage offered by these products in a parameter regionalization method that supports sub-grid variability. This study assessed the transferability of optimized model parameters from a gauged to an ungauged basin using the mesoscale Hydrologic Model (mHM)—Multiscale Parameter Regionalization (MPR) technique. The ability of the fifth generation European Centre for Medium-Range Weather Forecasts Reanalysis product (ERA5), Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS), Global Precipitation Climatology Centre (GPCC), and Multi-Source Weighted-Ensemble Precipitation (MSWEP) gridded rainfall products to simulate observed discharge in three basins was first assessed. Thereafter, the CHIRPS rainfall product was used in three multi-basin mHM setups. Optimized model parameters were then transferred to independent basins, and the reproduction of observed discharges was assessed. Kling–Gupta Efficiency (KGE) scores showed improvements when mHM runs were performed using optimized parameters in comparison to using default parameters for discharge simulations. Optimized mHM runs performed reasonably (KGE > 0.4) for all basins and rainfall products. However, only one basin showed a satisfactory KGE value (KGE = 0.54) when optimized parameters were transferred to an ungauged basin. This study underscores the utility of the mHM-MPR tool for parameter transferability during discharge simulation in data-scarce regions.
... Besides the process complexity, the spatial complexity of the model may also be an important factor that should be considered in the model. Natural systems always exhibit strong spatial heterogeneity introduced by geology, soil types, vegetation, or topography (Das et al. 2008;Kumar et al. 2010;Smith et al. 2004). Therefore, the hydrologic models should consider the spatial heterogeneous of the catchment to represent a more realistic model structure. ...
... Several studies indicated that the lumped model is versatile enough to represent the spatial heterogeneity of the catchment and it could provide very similar or even better performance than the distributed model (Ghavidelfar et al. 2011;Reed et al. 2004;Refsgaard and Knudsen 1996; Vansteenkiste et al. 2014). However, many other studies indicated that the model performance on the streamflow could be improved by the consideration of spatial complexity (Yan and Zhang 2014;Atkinson et al. 2003;Boyle et al. 2001b;Euser et al. 2015;Han et al. 2014;Kumar et al. 2010;Nijzink et al. 2016). Therefore, whether the spatial heterogeneity is needed in the hydrological model and what is the level of spatial complexity required in the model is still unclear and will depend on the system being studied. ...
Multi-model frameworks are widely used to identify the appropriate model structure for the study catchment. However, most frameworks mainly consider the process complexity of the model, and few of them consider the spatial complexity. In this paper, we investigated the appropriate model structure for a karst catchment from the aspect of spatial complexity. The purpose is twofold: (1) to investigate whether the spatial complexity is needed to simulate the spring discharge of this karst catchment and (2) to investigate whether the increase of model’s spatial complexity can make up its deficiency on the process complexity. Three simple lumped models with different process complexities were chosen to gradually increase the spatial heterogeneity of their parameters to investigate the appropriate model structure for simulating the discharge of a karst spring. The results show that the performances of three lumped models highly improve when adding the routing function to them. However, further considering the spatial parameter heterogeneity, only one model shows obvious performance improvement and other two models show limited improvement. Moreover, this model with relatively complex spatial parameter heterogeneity still shows worse performance than another lumped model. This indicates an increase of models’ spatial complexity cannot always make up their process deficiencies. The final comparison results indicated that the lumped model or their semi-lumped version with flexible process complexity is enough to simulate the discharge of this karst spring and no extra spatial complexity is needed. Our studies also indicated that the increase in spatial complexity of the model cannot always fully compensate its deficiency in process complexity.
... In spatially distributed models, free parameters have to be inferred through calibration for each modeling unit, thus increasing the number of parameters if the model resolution is increased [23,24]. One technique for reducing the number of free parameters is the Hydrological Response Unit (HRU) approach, applied, among others, by SWAT, where modeling units of the same physical characteristics (soil, land use, slope, etc.) are first grouped and then calibrated together [14,25,26]. Another method of parameter reduction, which is applied by the mesoscale Hydrologic Model (mHM) [18,21], is multiscale parameter regionalization (MPR) [18]. ...
... Multiple studies have explored the performance of the mHM model, mostly for the European continent [18,21,[26][27][28][29][30][31][32]. Kauffeldt et al. [33] give an overview of 24 large-scale hydrologic models and their suitability for the European Flood Awareness System. ...
The prediction of freshwater resources remains a challenging task in West Africa, where the decline of in situ measurements has a detrimental effect on the quality of estimates. In this study, we establish a series of modeling routines for the grid-based mesoscale Hydrologic Model (mHM) using Multiscale Parameter Regionalization (MPR). We provide a computationally efficient application of mHM-MPR across a diverse range of data-scarce basins using in situ observations, remote sensing, and reanalysis inputs. Model performance was first screened for four precipitation datasets and three evapotranspiration calculation methods. Subsequently, we developed a modeling framework in which the pre-screened model is first calibrated using discharge as the observed variable (mHM Q), and next calibrated using a combination of discharge and actual evapotranspiration data (mHM Q/ET). Both model setups were validated in a multi-variable evaluation framework using discharge, actual evapotranspiration, soil moisture and total water storage data. The model performed reasonably well, with mean discharge KGE values of 0.53 (mHM Q) and 0.49 (mHM Q/ET) for the calibration; and 0.23 (mHM Q) and 0.13 (mHM Q/ET) for the validation. Other tested variables were also within a good predictive range. This further confirmed the robustness and well-represented spatial distribution of the hydrologic predictions. Using MPR, the calibrated model can then be scaled to produce outputs at much smaller resolutions. Overall, our analysis highlights the worth of utilizing additional hydrologic variables (together with discharge) for the reliable application of a distributed hydrologic model in sparsely gauged West African river basins.
... In this study, we rigorously test SPAEF and compare it with two additional spatial performance metrics; namely fractions skill score (Roberts and Lean, 2008) and connectivity analysis (Koch et al., 2016b). All three metrics are applied in a spatial pattern oriented calibration of a catchment model using the multiscale Hydrologic Model (mHM: Samaniego et al., 2010). 20 ...
... This study utilizes the mesoscale Hydrologic Model (mHM; version 5.6) which is a grid based spatially distributed hydrological model (Kumar et al., 2013, 2010, Samaniego et al., 2010a, 2010b. The model accounts for key hydrological 5 processes such as canopy interception, soil moisture dynamics, surface/subsurface flow generation, snow melting, This was achieved by adding effective calibration parameters to the soil moisture stress function, root fraction coefficient and the dynamic scaling of reference ET by incorporating the Moderate Resolution Imaging Spectroradiometer (MODIS) 8day Leaf Area Index (LAI) product at 1 km 2 resolution. ...
The process of model evaluation is not only an integral part of model development and calibration but also of paramount importance when communicating modelling results to the scientific community and stakeholders. The modelling 10 community has a large and well tested toolbox of metrics to evaluate temporal model performance. On the contrary, spatial performance evaluation is not corresponding to the grand availability of spatial observations readily available and to the sophisticate model codes simulating the spatial variability of complex earth system processes. This study makes a contribution towards advancing spatial pattern oriented model calibration by rigorously testing a multiple-component performance metric. The promoted SPAtial EFficiency (SPAEF) metric reflects three equally weighted components: 15 correlation, coefficient of variation and histogram overlap. This multiple-component approach is found to be advantageous in order to achieve the complex task of comparing spatial patterns. SPAEF, its three components individually and two alternative spatial performance metrics, i.e. connectivity analysis and fractions skill score, are applied in a spatial pattern oriented model calibration of a catchment model in Denmark. Results suggest the importance of multiple-component metrics, because stand-alone metrics tend to fail to provide holistic pattern information to the optimizer. The three SPAEF 20 components are found to be independent which allows them to complement each other in a meaningful way. In order to optimally exploit spatial observations made available by remote sensing platforms this study suggests applying bias insensitive metrics which further allow comparing variables which are related but may differ in unit. This study applies SPAEF in the hydrological context using the mesoscale Hydrologic Model (mHM; version 5.6), but we see great potential across disciplines related to spatial distributed earth system modelling. 25
... Poff et al., 1997;Poff et al., 2010). Five main characteristics of flow regime are to be considered: magnitude, timing/seasonality, frequency, duration, and rate of change Poff et al., 2010;Kumar et al., 2010;Archfield et al., 2014;Laizé et al., 2014). Almost two hundred Indicators of Hydrologic Alteration (IHAs), each assessing one or more flow regime characteristics, have been used in flow-ecological assessments. ...
... Uncertainty of mean annual flow has been estimated at about ±10-15%, but it may exceed ±20% at low or high flow percentiles. Uncertainty for indicators of frequency and duration of high and low flow is even higher, in the range of ±30-40%, especially if the indicators are defined in relation to a threshold (e.g. as a multiple of mean or median flow or flows crossing specified quantile levels; Kumar et al., 2010;Westerberg and McMillan, 2015;Westerberg et al., 2016). Kennard et al. (2010) analyzed the impact of the period length on the accuracy of 120 IHAs and found that indicator accuracy quickly improved when the period of analysis increases from one to 15 years, but after that indicators tended to stabilize, and did not change substantially for periods longer than 30 years. ...
Abstract Sustainable water basin management requires characterization of flow regime in river networks impacted by anthropogenic pressures. Flow regime in ungauged catchments under current, future, or natural conditions can be assessed with hydrological models. Developing hydrological models is, however, resource demanding such that decision makers might revert to models that have been developed for other purposes and are made available to them (‘off-the-shelf’ models). In this study, the impact of epistemic uncertainty of flow regime indicators on flow-ecological assessment was assessed at selected stations with drainage areas ranging from about 400 to almost 90,000 km2 in four South European basins (Adige, Ebro, Evrotas and Sava). For each basin, at least two models were employed. Models differed in structure, data input, spatio-temporal resolution, and calibration strategy, reflecting the variety of conditions and purposes for which they were initially developed. The uncertainty of modelled flow regime was assessed by comparing the modelled hydrologic indicators of magnitude, timing, duration, frequency and rate of change to those obtained from observed flow. The results showed that modelled flow magnitude indicators at medium and high flows were generally reliable, whereas indicators for flow timing, duration, and rate of change were affected by large uncertainties, with correlation coefficients mostly below 0.50. These findings mirror uncertainty in flow regime indicators assessed with other methods, including from measured streamflow. The large indicator uncertainty may significantly affect assessment of ecological status in freshwater systems, particularly in ungauged catchments. Finally, flow-ecological assessments proved very sensitive to reference flow regime (i.e., without anthropogenic pressures). Model simulations could not adequately capture flow regime in the reference sites comprised in this study. The lack of reliable reference conditions may seriously hamper flow-ecological assessments. This study shows the pressing need for improving assessment of natural flow regime at pan-European scale.
... There have been many attempts to improve the parameterization of lumped and semi-distributed models by further discretizing the sub-basins into a given number of regions that exhibit nearly similar hydrologic behavior, i.e., the so-called HRU concept initially proposed by Leavesley et al. (1983) and further developed by others (e.g., Flügel, 1995;Beldring et al., 2003;Blöschl et al., 2008;Viviroli et al., 2009;Zehe et al., 2014). Unfortunately, results obtained in these parameterization attempts have not been very successful in realistically representing the spatial variability of model parameters, states, and fluxes because of the lack of regionalized parameters and the unabridged reliance on parameter calibration to improve model performance (Kumar et al., 2010). Commonly, the effective parameters estimated for the HRUs are found by automatic calibration. ...
... Efforts have been made to enforce continuity on parameter fields (Gotzinger and Bárdossy, 2007;Singh et al., 2012) but with somewhat limited success during the transferability of parameters across scales and locations. In addition, models parameterized using HRUs do not lead to mass conservation of water fluxes (i.e., flux-matching) when applied to scales other than those used for calibration (Kumar et al., 2010(Kumar et al., , 2013b. Recent attempts have been made to improve the HRU concept to increase the seamless representation of parameters, states, and fluxes (Chaney et al., 2016a). ...
Land surface and hydrologic models (LSMs/HMs) are used at diverse
spatial resolutions ranging from catchment-scale (1–10 km) to
global-scale (over 50 km) applications. Applying the same model
structure at different spatial scales requires that the model
estimates similar fluxes independent of the chosen resolution, i.e.,
fulfills a flux-matching condition across scales. An analysis of
state-of-the-art LSMs and HMs reveals that most do not have consistent
hydrologic parameter fields. Multiple experiments with the mHM,
Noah-MP, PCR-GLOBWB, and WaterGAP models demonstrate the pitfalls of
deficient parameterization practices currently used in most
operational models, which are insufficient to satisfy the
flux-matching condition. These examples demonstrate that J. Dooge's
1982 statement on the unsolved problem of parameterization in these
models remains true. Based on a review of existing parameter
regionalization techniques, we postulate that the multiscale parameter
regionalization (MPR) technique offers a practical and robust method
that provides consistent (seamless) parameter and flux fields across
scales. Herein, we develop a general model protocol to describe how
MPR can be applied to a particular model and present an example
application using the PCR-GLOBWB model. Finally, we discuss potential
advantages and limitations of MPR in obtaining the seamless prediction
of hydrological fluxes and states across spatial scales.
... This is sometimes referred to as semi-distributed approach, often being the preferred choice in practical applications (e.g. Kumar et al., 2010;Euser et al., 2015). Although no clear definition exists, this family of discretisation typically consists of a hierarchical multi-scale scheme dividing a hydrological basin into several subbasins, which in turn contain irregularly shaped computational elements being hydrologically uniform entities (e.g. ...
... The role of detail of discretisation (i.e. the spatial resolution) of hydrological models, partly also as a user decision during the pre-processing process, has long been acknowledged in numerous studies (e.g. Wood et al., 1988;Kumar et al., 2010;Han et al., 2014;Euser et al., 2015;Haghnegahdar et al., 2015;González et al., 2016). For gridbased models, this influence has been thoroughly assessed (e.g. ...
The characteristics of a landscape pose essential factors for
hydrological processes. Therefore, an adequate representation of the
landscape of a catchment in hydrological models is vital. However, many of
such models exist differing, amongst others, in spatial concept and
discretisation. The latter constitutes an essential pre-processing step, for
which many different algorithms along with numerous software implementations
exist. In that context, existing solutions are often model specific,
commercial, or depend on commercial back-end software, and allow only a
limited or no workflow automation at all.
Consequently, a new package for the scientific software and scripting
environment R, called lumpR, was developed. lumpR employs an
algorithm for hillslope-based landscape discretisation directed to
large-scale application via a hierarchical multi-scale approach. The package
addresses existing limitations as it is free and open source, easily
extendible to other hydrological models, and the workflow can be fully
automated. Moreover, it is user-friendly as the direct coupling to a GIS
allows for immediate visual inspection and manual adjustment. Sufficient control
is furthermore retained via parameter specification and the option to include
expert knowledge. Conversely, completely automatic operation also allows for
extensive analysis of aspects related to landscape discretisation.
In a case study, the application of the package is presented. A sensitivity
analysis of the most important discretisation parameters demonstrates its
efficient workflow automation. Considering multiple streamflow metrics, the
employed model proved reasonably robust to the discretisation parameters.
However, parameters determining the sizes of subbasins and hillslopes proved
to be more important than the others, including the number of representative
hillslopes, the number of attributes employed for the lumping algorithm, and
the number of sub-discretisations of the representative hillslopes.
... There have been many attempts to improve the parameterization of lumped and semi-distributed models by further discretizing the sub-basins into a given number of regions that exhibit nearly similar hydrologic behavior, i.e., the so-called HRU concept initially proposed by Leavesley et al. (1983) and further developed by others (e.g., Flügel, 1995;Beldring et al., 2003;Blöschl et al., 2008;Viviroli et al., 2009;Zehe et al., 2014). Unfortunately, results obtained in these parameterization attempts have not been very successful in realistically representing the spatial variability of model parameters, states, and fluxes because of the lack of regionalized parameters and the unabridged reliance on parameter calibration to improve model performance (Kumar et al., 2010). Commonly, the effective parameters estimated for the HRUs are found by automatic calibration. ...
... Efforts have been made to enforce continuity on parameter fields (Gotzinger and Bárdossy, 2007;Singh et al., 2012) but with somewhat limited success during the transferability of parameters across scales and locations. In addition, models parameterized using HRUs do not lead to mass conservation of water fluxes (i.e., flux-matching) when applied to scales other than those used for calibration (Kumar et al., 2010(Kumar et al., , 2013b. Recent attempts have been made to improve the HRU concept to increase the seamless representation of parameters, states, and fluxes (Chaney et al., 2016a). ...
Land surface and hydrologic models (LSM/HM) are used at diverse spatial resolutions ranging from 1–10 km in catchment-scale applications to over 50 km in global-scale applications. Application of the same model structure at different spatial scales requires that the model estimates similar fluxes independent of the model resolution and fulfills a flux-matching condition across scales. An analysis of state-of-the-art LSMs and HMs reveals that most do not have consistent and realistic parameter fields for land surface geophysical properties. Multiple experiments with the mHM, Noah-MP, PCR-GLOBWB and WaterGAP models are conducted to demonstrate the pitfalls of poor parameterization practices currently used in most operational models, which are insufficient to satisfy the flux-matching condition. These examples demonstrate that J. Dooge's 1982 statement on the unsolved problem of parameterization in these models remains true. We provide a short review of existing parameter regionalization techniques and discuss a method for obtaining seamless hydrological predictions of water fluxes and states across multiple spatial resolutions. The multiscale parameter regionalization (MPR) technique is a practical and robust method that provides consistent (seamless) parameter and flux fields across scales. A general model protocol is presented to describe how MPR can be applied to a specific model, with an example of this application using the PCR-GLOBWB model. Applying MPR to PCR-GLOBWB substantially improves the flux-matching condition. Estimation of evapotranspiration without MPR at 5 arcmin and 30 arcmin spatial resolutions for the Rhine river basin results in a difference of approximately 29 %. Applying MPR reduce this difference to 9 %. For total soil water, the differences without and with MPR are 25 % and 7 %, respectively.
... In other to overcome the problems of over-parameterization and equifinality, the Mesoscale Parameter Regionalization (MPR) proposed by Samaniego et al. (2010) presents a technique which links model parameters at a coarser scale with their counterparts at a finer resolution using pedotransfer functions, whereby only the global parameters that define these relationships are obtained through calibration. In comparison with other regionalization methods (e.g. standard regionalization), MPR showed superiority in preserving the spatial variability of state variables and overall performance of model hydrologic processes simulation Kumar et al., 2010). The MPR technique reduces the number of free mHM calibration parameters and seeks to address Question 20 (reducing model uncertainty) of the Unsolved Problems in Hydrology (UPH 20) (Blöschl et al., 2019). ...
Predictive hydrologic modelling to understand and support agricultural water resources management and food security policies in Nigeria is a demanding task due to the paucity of hydro-meteorological measurements. This study assessed the skill of using different remotely sensed rainfall products in a multi-calibration framework for evaluating the performance of the mesoscale hydrologic Model (mHM) across four different data-scarce basins in Nigeria. Grid-based rainfall estimates obtained from several sources were used to drive the mHM in different basins in Nigeria. Model calibration was first performed using only discharge records, and also by using a combination of discharge and actual evapotranspiration, forced with different rainfall products. The mHM forced with CHIRPS produced reasonable Kling-Gupta efficiency KGE) results (0.5> KGE <0.85) under both calibration frameworks. However, constraining model parameters under a multi-calibration arrangement showed no significant discharge simulation improvement in this study. Results show the utility of the mHM for discharge simulation in data-sparse basins in Nigeria.
... It is important to consider the differences between the compared resolutions in such studies (Petrovic et al., 2022). Besides the model resolution, the model setup (the domain configuration and physical parameterizations for the selected target region) is a crucial factor for reliable simulations (e.g., Stoelinga et al., 2003;Kumar et al., 2010). For the temperature simulation, Vautard et al. (2013) found that it is primarily sensitive to convection and microphysics schemes. ...
Heat waves are among the most severe climate extreme events. In this study, we address the impact of increased model resolution and tailored model settings on the reproduction of these events by evaluating different regional climate model outputs for Germany and its near surroundings between 1980–2009. Outputs of an ensemble of six EURO-CORDEX models with 12.5 km grid resolution and outputs from a high-resolution (5 km) WRF (Weather Research and Forecasting) model run are employed. The latter was especially tailored for the study region regarding the physics configuration. We analyze the reproduction of the maximum temperature, number of heat wave days, heat wave characteristics (frequency, duration and intensity), the 2003 major event, and trends in the annual number of heat waves. E-OBS is used as the reference, and we utilize the Taylor diagram, the Mann–Kendall trend test and the spatial efficiency metric, while the cumulative heat index is used as a measure of intensity. Averaged over the domain, heat waves occurred about 31 times in the study period, with an average duration of 4 d and an average heat excess of 10 ∘C. The maximum temperature was only reproduced satisfactorily by some models. Despite using the same forcing, the models exhibited a large spread in heat wave reproduction. The domain mean conditions for heat wave frequency and duration were captured reasonably well, but the intensity was reproduced weakly. The spread was particularly pronounced for the 2003 event, indicating how difficult it was for the models to reproduce single major events. All models underestimated the spatial extent of the observed increasing trends. WRF generally did not perform significantly better than the other models. We conclude that increasing the model resolution does not add significant value to heat wave simulation if the base resolution is already relatively high. Tailored model settings seem to play a minor role. The sometimes pronounced differences in performance, however, highlight that the choice of model can be crucial.
... The influence of different discretization methods on hydrologic responses has been acknowledged in multitudes of studies (Kumar et al. 2010;Euser et al. 2015;González et al. 2016;Caldeira et al. 2019). However, this influence has been assessed extensively in grid-based models (Molnar and Julien 2000;Sulis et al. 2011;Euser et al. 2015;Melsen et al. 2016) as it is relatively easy to change the grid resolution. ...
Accurate streamflow simulation is crucial for effective hydrologic forecasting and water resource management. This study introduces a nested discretization scheme aimed at refining catchment delineation based on the spatial heterogeneity of its characteristics. The scheme aims to align with the assumption of spatial homogeneity within a hydrologic model, enhancing simulation accuracy. Investigating the impact of discretization, the study evaluates lumped, semi-lumped, and semi-distributed conceptual model structures, both in continuous and event-based simulations of flood events in Jagdalpur and Wardha basins of India. Results indicate superior performance by continuous semi-distributed and semi-lumped models (efficiency > 0.77), followed by continuous lumped models (efficiency > 0.68) during both calibration and validation periods at both basins. Event-based models, particularly semi-distributed and semi-lumped, exhibit higher median efficiency (> 0.71 at Jagdalpur and > 0.67 at Wardha) compared to their lumped counterparts (0.57 at Jagdalpur and 0.27 at Wardha), showcasing their proficiency in capturing spatial variability. However, a marginal performance increase in semi-lumped models with increased spatial discretization is observed, accompanied by a significant rise in computational time. This research contributes insights into the trade-offs associated with the proposed discretization scheme and emphasizes the balance between model complexity and efficiency for optimal streamflow simulations.
... It is important to consider the differences between the compared resolutions in such studies (Petrovic et al., 2022). Besides the model resolution, the model setup regarding the domain configuration and physical parameterizations for the selected target region is a crucial factor for reliable simulations (e.g., Stoelinga et al., 2003;Kumar et al., 2010). For the temperature simulation, Vautard et al. (2013) found that it is primarily sensitive to convection and microphysics schemes. ...
Heat waves are among the most severe climate extreme events. In this study, we address the impact of increased model resolution and tailored model settings on the reproduction of these events by evaluating different regional climate model outputs for Germany and the near surroundings between 1980–2009. Both, outputs of an ensemble of six EURO-CORDEX models of 12.5 km grid resolution and outputs from a high resolution (5 km) WRF run are employed. The latter was especially tailored for the study region regarding the physics configuration. We analyze the reproduction of maximum temperature, number of heat wave days, heat wave characteristics (frequency, duration and intensity), the 2003 major event and trends in the annual number of heat waves. E-OBS is used as reference and we imply Taylor diagram, Mann-Kendall trend test, spatial efficiency metric and cumulative heat index as a measure for intensity. Averaged over the domain, heat waves occurred about 31 times in the study period with an average duration of 4 days and average heat excess of 10 °C. Maximum temperature was reproduced reasonably well by all models. Despite the same forcing, the models exhibited a large spread in the heat wave reproduction. The domain mean conditions of heat wave frequency and duration were captured reasonably well, but intensity was reproduced weakly. The spread was particularly pronounced for the 2003 event, indicating the difficulty of models to reproduce single major events. All models underestimated the spatial extent of the observed increasing trends. WRF mostly did not perform significantly better than the other models. We conclude that increased model resolution does not add a significant value to heat wave simulation if the base resolution is already relatively high. Tailored model settings seem to play a minor role. The partly distinct differences in performance, however, highlight that the choice of model can be crucial.
... This can improve the representation of spatial structures (Fohrer et al. 2016). It is commonly used for interpolating precipitation and temperature data on a catchment scale (Grayson & Blöschl 2001;Das et al. 2008;Kumar et al. 2010). Elevation was used as an external drift since there is a high correlation between elevation with precipitation and temperature. ...
Data scarcity in many areas around the world represents a major problem for hydrological model calibrations. Global parameter estimates and global forcing can provide possibilities to access hydrological responses in ungauged regions. In this study, we applied HBV global parameter estimates considering uncertainty in the Upper Neckar and Upper Danube catchments, Germany, to answer what are the influencing factors and how good are their local applications. We tested simulations with precipitation in spatial resolutions from 0.05 to 0.2° and with local/global sources. Results show that the general performance is acceptable to good (Kling-Gupta efficiency, KGE: 0.51–0.79) in both catchments using local or global precipitation. The influence of spatial resolutions is insignificant while using local precipitation slightly increases performance in both catchments. Catchment properties such as complex topography and special karst subsurface may lead to a deterioration of performance by 0.2 of median KGE in the Upper Danube compared with the Upper Neckar catchment. The median correlation coefficient, runoff ratio and relative error suggest that using global parameter estimates can reproduce seasonality and long-term water balance in our studied region. Our study highlights the potential of using global parameter estimates and global forcing in ungauged areas.
HIGHLIGHTS
Global parameter estimates with global forcing can produce acceptable local discharge simulations in our study area.;
Catchment properties have more impacts on local applications than sources and spatial resolutions of forcing when applying global parameter estimates.;
Our study highlights the potential of using global forcing to understand hydrological behavior and water balance at a monthly scale in ungauged areas.;
... The mesoscale Hydrological Model (mHM) was developed by UFZ (Helmholtz Centre for Environmental Research) [21,22]. mHM is a fully distributed, physically based, and continuous model. ...
The spatial heterogeneity in hydrologic simulations is a key difference between lumped and distributed models. Not all distributed models benefit from pedo-transfer functions based on the soil properties and crop-vegetation dynamics. Mostly coarse-scale meteorological forcing is used to estimate only the water balance at the catchment outlet. The mesoscale Hydrologic Model (mHM) is one of the rare models that incorporate remote sensing data, i.e., leaf area index (LAI) and aspect, to improve the actual evapotranspiration (AET) simulations and water balance together. The user can select either LAI or aspect to scale PET. However, herein we introduce a new weight parameter, "alphax", that allows the user to incorporate both LAI and aspect together for potential evapotranspiration (PET) scaling. With the mHM code enhancement, the modeler also has the option of using raw PET with no scaling. In this study, streamflow and AET are simulated using the mHM in The Main Basin (Germany) for the period of 2002-2014. The additional value of PET scaling with LAI and aspect for model performance is investigated using Moderate Resolution Imaging Spectroradiometer (MODIS) AET and LAI products. From 69 mHM parameters, 26 parameters are selected for calibration using the Optimization Software Toolkit (OSTRICH). For calibration and evaluation, the KGE metric is used for water balance, and the SPAEF metric is used for evaluating spatial patterns of AET. Our results show that the AET performance of the mHM is highest when using both LAI and aspect indicating that LAI and aspect contain valuable spatial heterogeneity information from topography and canopy (e.g., forests, grasslands, and croplands) that should be preserved during modeling. This is key for agronomic studies like crop yield estimations and irrigation water use. The additional "alphax" parameter makes the model physically more flexible and robust as the model can decide the weights according to the study domain.
... Therefore, it is necessary to analyze the impacts of both ENSO and IOD events on the temporal and spatial patterns of regional streamflow in the JRB. Reportedly, extreme precipitation events also play an important role in flood control and ecology protection [29,30], implying that the effects of ENSO and IOD events on extreme precipitation conditions should also be investigated. Additionally, some studies analyzed the impacts of ENSO and IOD events throughout the period of their occurrence [4,31]; however, the results were unclear because the lag effect of the climate anomalies was not fully considered [27]. ...
This study investigated the combined impacts of the El Niño–Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD) on streamflow under four scenarios: neutral, pure ENSO, pure IOD, and a combination of ENSO and IOD. The Jinsha River Basin (JRB), at the head of the Yangtze River, was used as a case study. By using statistical methods such as coherent wavelet analysis (WTC), we are committed to studying what kind of impact the IOD will have, the difference in impact between ENSO and IOD at different stages, and the difference in impact between ENSO and IOD on the mean and extreme values of runoff, compared with traditional single ENSO event, to provide support for water resource management, especially for reservoir operation. The key results are as follows. (a) Both ENSO and IOD events affect annual and seasonal streamflow in the JRB. (b) The impact of pure IOD events on annual streamflow in the JRB was twice as great as that of pure ENSO events in developing years, whereas the opposite was true in decaying years. (c) The combined impact of ENSO and IOD led to a higher streamflow maximum than the annual or seasonal average streamflow. Conversely, their impact on the streamflow minima was less than 10% during both developing and decaying years, except at Zhimenda Station. (d) Overall, water shortages could be more serious in developing years than in neutral years, and much more attention should be given to flooding control in decaying years. These results can be used as a reference for water resource management concerning agricultural planning and ecological protection in the JRB.
... In the recent past, Germany and other parts of central Europe have been hit by dryness in the summer periods. Especially the severe drought events in 2015 (e.g., Hoy et al., 2017;Ionita et al., 2017;Laaha et al., 2017), 2018 (e.g., Bastos et al., 2020;Thompson et al., 2020) and 2019 (e.g., Boergens et al., 2020;Hari et al., 2020;Ziernicka-Wojtaszek, 2021), which occurred in combination with heat waves, have contributed to this. In addition, 2020 was also categorized as too dry, mainly in the spring and summer months (DWD, 2020;Umweltbundesamt, 2021). ...
Droughts are among the most relevant natural disasters related to climate change. We evaluated different regional climate model outputs and their ability to reproduce observed drought indices in Germany and its near surroundings between 1980–2009. Both outputs of an ensemble of six EURO-CORDEX models of 12.5 km grid resolution and outputs from a high-resolution (5 km) Weather Research and Forecasting (WRF) run were employed. The latter model was especially tailored for the study region regarding the physics configuration. We investigated drought-related variables and derived the 3-month standardized precipitation evapotranspiration index (SPEI-3) to account for meteorological droughts. Based on that, we analyzed correlations, the 2003 event, trends and drought characteristics (frequency, duration and severity) and compared the results to E-OBS. Methods used include Taylor diagrams, the Mann–Kendall trend test and the spatial efficiency (SPAEF) metric to account for spatial agreement of patterns. Averaged over the domain, meteorological droughts were found to occur approximately 16 times in the study period with an average duration of 3.1 months and average severity of 1.47 SPEI units. WRF's resolution and setup were shown to be less important for the reproduction of the single drought event and overall drought characteristics. Depending on the specific goals of drought analyses, computation resources could therefore be saved, since a coarser resolution can provide similar results. Benefits of WRF were found in the correlation analysis. The greatest benefits were identified in the trend analysis: only WRF was able to reproduce the observed negative SPEI trends to a fairly high spatial accuracy, while the other regional climate models (RCMs) completely failed in this regard. This was mainly due to the WRF model settings, highlighting the importance of appropriate model configuration tailored to the target region. Our findings are especially relevant in the context of climate change studies, where the appropriate reproduction of trends is of high importance.
... These models are prevalent and practical tools for estimating hydro-sedimentological fluxes and sources under current conditions and possible future scenarios. However, a watershed model's fidelity hinges on various factors (Kumar et al., 2010), of which the quality of the model calibration is one of the critical factors (Molina-Navarro et al., 2017). Minimizing compensation error while enhancing simulation performance (Ahmadi et al., 2014) is an important part of a good calibration technique that improves model reliability in solving practical issues (Gupta et al. 1998;1999). ...
Spatially distributed watershed models are commonly utilized to address a wide range of water-related issues. However, setting up a reliable watershed model is a difficult task involving several essential decisions making. Choice of calibration method is one of the most important decisions that has been sparsely investigated in semi-distributed watershed models. In this study, therefore, we used the Soil and Water Assessment Tool (SWAT) model to investigate the impact of three calibration methods: sequential (SQN), simultaneous (SML) and sequential-simultaneous (SQN_SML) on model performance and parameter uncertainty in the Kantamal catchment of the Mahanadi basin, India. The findings across the calibration methods; evaluated fit scores of streamflow for respective calibration and validation period; showed that SQN_SML calibration has the least amount of bias (PBAIS = 1.7, -4.2), the highest NSE (0.91, 0.92), KGE (0.95, 0.94) and R² (0.91, 0.92). Furthermore, SQN_SML outperformed the other two methods in all three streamflow regimes (low, medium and high) of flow duration curve analysis. Suspended sediment load (SSL) analyses of partitioned sediment duration curve showed the best performance of SQN_SML for mid and low SSL regimes while all three calibration methods performed similarly in the high SSL regime. SML calibration approach showed the least parameter uncertainty followed by SQN_SML and SQN. The P-factor for sediment simulation was better for the SQN_SML approach, indicating the minimal model error for sediment simulation. The SQN_SML produced the least equifinal solution, while the SQN approach produced the highest equifinal solution. Overall, the findings of this study may help the watershed modelling communities for selecting suitable calibration strategies when dealing with integrated water resources management.
... continuous internal change (e.g., land use change) and boundary conditions (e.g., changing climate), distributed hydrological models have been used across different spatio-temporal scales (Addor et al., 2014;Blöschl et al., 2008;Famiglietti & Wood, 1995;Kumar et al., 2010Kumar et al., , 2013Martel et al., 2020;Merz & Blöschl, 2004;Rakovec et al., 2016;Thober et al., 2019;Wanders & Wada, 2015). However, the models themselves suffer from inadequate simulation of hydrological processes due to a lack of scale-relevant theories in watershed hydrology (Blöschl & Sivapalan, 1995;Dooge, 1986;Peters-Lidard et al., 2017;Samaniego et al., 2017). ...
Quantifying the uncertainty linked to the degree to which the spatio‐temporal variability of the catchment descriptors (CDs), and consequently calibration parameters (CPs), represented in the distributed hydrology models and its impacts on the simulation of flooding events is the main objective of this paper. Here, we introduce a methodology based on ensemble approach principles to characterize the uncertainties of spatio‐temporal variations. We use two distributed hydrological models (WaSiM and Hydrotel) and six catchments with different sizes and characteristics, located in southern Quebec, to address this objective. We calibrate the models across four spatial (100, 250, 500, 1000 m2) and two temporal (3 hours and 24 hours) resolutions. Afterwards, all combinations of CDs‐CPs pairs are fed to the hydrological models to create an ensemble of simulations for characterizing the uncertainty related to the spatial resolution of the modeling, for each catchment. The catchments are further grouped into large (>1000 km2), medium (between 500 and 1000 km2) and small (<500 km2) to examine multiple hypotheses. The ensemble approach shows a significant degree of uncertainty (over 100% error for estimation of extreme streamflow) linked to the spatial discretization of the modeling. Regarding the role of catchment descriptors, results show that first, there is no meaningful link between the uncertainty of the spatial discretization and catchment size, as spatio‐temporal discretization uncertainty can be seen across different catchment sizes. Second, the temporal scale plays only a minor role in determining the uncertainty related to spatial discretization. Third, the more physically representative a model is, the more sensitive it is to changes in spatial resolution. Finally, the uncertainty related to model parameters is larger than that of catchment descriptors for most of the catchments. Yet, there are exceptions for which a change in spatio‐temporal resolution can alter the distribution of state and flux variables, change the hydrologic response of the catchments, and cause large uncertainties. This article is protected by copyright. All rights reserved.
... The most commonly used parameterization technique is to group homogeneous regions using basin physical characteristics, and to identify a unique set of parameters for these groups through calibration. This approach is performed in models in which the process representations and computations are done in hydrologically divided sub-basins (Das et al. 2008;Kumar et al. 2010;Shi et al. 2013). ...
Distributed hydrological models account for spatial heterogeneity by discretizing the watershed into unique units based on the watershed characteristics. However, parameter estimation is one of the major tasks in the application of distributed hydrological models. The existing calibration methods for distributed hydrological models do not consider the spatial variability of the parameters across the basin, and, therefore, do not guarantee good simulations on locations other than the calibration outlets. This study proposes a calibration approach which preserves the heterogeneity of the parameters across the basin. The basic simulation units of the distributed models are grouped in this approach based on the land use and soil type, and a random perturbation of the parameters is performed in these groups during calibration. The proposed method is demonstrated through a case study of two watersheds in the USA using soil and water assessment tool (SWAT) model. The results indicate that the calibrated model simulations in the upstream gauged locations (other than that used for calibration) are much better in the proposed approach, in contrast to the currently employed calibration method. Nonetheless, it is also observed that the proposed calibration approach would be more effective in watersheds that have higher spatial heterogeneity.
... However, the compared models differ in their approach to spatial discretization (i.e., VIC is grid-based whereas PRMS and SUMMA use hydrologic response units), and their study relies on only four watersheds. Previous studies have indicated that the HM output is sensitive to the HM spatial discretization (Kumar et al., 2010;Sciuto & Diekkrüger, 2010). Also, testing TE over a wide variety of catchments would establish relationships between process connectivity and catchment characteristics to provide insights into more detailed hydrologic behaviors (Götzinger & Bárdossy, 2007;Oudin et al., 2010;Sivapalan, 2003). ...
Exploring water fluxes between hydrological model (HM) components is essential to assess and improve model realism. Many classical metrics for HM diagnosis rely solely on streamflow and hence provide limited insights into model performance across processes. This study applies an information theory measure known as "transfer entropy" (TE) to systematically quantify the transfer of information among major HM components. To test and demonstrate the benefits of TE, we use the Framework for Understanding Structural Errors (FUSE) model to mimic and compare four commonly used HM structures, VIC, PRMS, SACRAMENTO, and TOPMODEL, across 671 catchments spanning a variety of hydrologic regimes in the conterminous United States. We explore connections between HM components and catchment landscape characteristics (e.g., climate, topography, soil, and vegetation) and characterize their nonlinear associations using distance correlation and Spearman correlation coefficients. Our results indicate that while the information transferred from precipitation to runoff is similar across model structures (likely as a result of calibration), the information transferred among other components can vary significantly from a FUSE structure to another. We find that aridity, precipitation duration and frequency, snow fraction, mean elevation, forest area, and leaf area index are often significantly associated with TE between the main HM components. We propose that the presence of meaningful nonlinear associations can be used to diagnose process representation in HMs. Our results highlight the necessity to enhance the conventional streamflow-only calibration approach for a more realistic representation of water dynamics in the models.
... However, the use of contrasting performance criteria that are focused on different hydrological aspects leads to a trade-off between them in optimising parameter values, since the selected best parameter values are likely to be different between these criteria (Kiesel et al. 2017). For instance, a parameter value which leads to a good representation of high flows can fail in reproducing low flows and vice versa (Madsen 2000, Boyle et al. 2001, Bekele and Nicklow 2007, Kumar et al. 2010, Zhang et al. 2011, Pfannerstill et al. 2014b. ...
Reliable simulations of hydrological models require that model parameters are precisely identified. In constraining model parameters to small ranges, high parameter identifiability is achieved. In this study, it is investigated how precisely model parameters can be constrained in relation to a set of contrasting performance criteria. For this, model simulations with identical parameter samplings are carried out with a hydrological model (SWAT) applied to three contrasting catchments in Germany (lowland, mid-range mountains, alpine regions). Ten performance criteria including statistical metrics and signature measures are calculated for each model simulation. Based on the parameter identifiability that is computed separately for each performance criterion, model parameters are constrained to smaller ranges individually for each catchment. An iterative repetition of model simulations with successively constrained parameter ranges leads to more precise parameter identifiability and improves model performance. Based on these results, a more consistent handling of model parameters is achieved for model calibration.
... The most widely used performance metrics are based on comparisons of simulated and observed response time series, including the mean squared error (MSE), Nash-Sutcliffe efficiency (NSE; a normalized version of MSE), and root mean squared error (RMSE; a transformation of MSE). Many previous studies have examined different variants of these metrics (e.g., see Oudin et al., 2006;Kumar et al., 2010;Pushpalatha et al., 2012;Price et al., 2012;Wöhling et al., 2013;Ding et al., 2016;Garcia et al., 2017), including their application to transformations of the system response time series to emphasize performance for specific flow regimes (e.g., use of logarithmic transformation to target low flows) or using combinations of different metrics to obtain balanced performance on different flow regimes. ...
Calibration is an essential step for improving
the accuracy of simulations generated using hydrologic models. A key modeling
decision is selecting the performance metric to be optimized. It has been
common to use squared error performance metrics, or normalized variants such
as Nash–Sutcliffe efficiency (NSE), based on the idea that their
squared-error nature will emphasize the estimates of high flows. However, we
conclude that NSE-based model calibrations actually result in poor
reproduction of high-flow events, such as the annual peak flows that are used
for flood frequency estimation. Using three different types of performance
metrics, we calibrate two hydrological models at a daily step, the Variable
Infiltration Capacity (VIC) model and the mesoscale Hydrologic Model (mHM),
and evaluate their ability to simulate high-flow events for 492 basins
throughout the contiguous United States. The metrics investigated are
(1) NSE, (2) Kling–Gupta efficiency (KGE) and its variants, and (3) annual
peak flow bias (APFB), where the latter is an application-specific metric
that focuses on annual peak flows. As expected, the APFB metric produces the
best annual peak flow estimates; however, performance on other
high-flow-related metrics is poor. In contrast, the use of NSE results in
annual peak flow estimates that are more than 20 % worse, primarily due
to the tendency of NSE to underestimate observed flow variability. On the
other hand, the use of KGE results in annual peak flow estimates that are
better than from NSE, owing to improved flow time series metrics (mean and
variance), with only a slight degradation in performance with respect to
other related metrics, particularly when a non-standard weighting of the
components of KGE is used. Stochastically generated ensemble simulations
based on model residuals show the ability to improve the high-flow metrics,
regardless of the deterministic performances. However, we emphasize that
improving the fidelity of streamflow dynamics from deterministically
calibrated models is still important, as it may improve high-flow metrics
(for the right reasons). Overall, this work highlights the need for a deeper
understanding of performance metric behavior and design in relation to the
desired goals of model calibration.
... Les données météorologiques constituant les données d'entrée du modèle hydrologique, l'élément qui les fournit n'en fait pas partie stricto sensu. Néanmoins, les caractéristiques de ces données (résolution spatio-temporelle, nature, qualité, et quantité) ont une influence déterminante sur le fonctionnement du modèle hydrologique, ce qui fait que ce dernier devra être construit de manière cohérente vis-à-vis des données météorologiques utilisées, particulièrement au niveau du découpage spatial et du niveau de représentation des processus (SCHAEFLI et al., 2005 ;SHRESTHA, TA-CHIKAWA et TAKARA, 2006 ;KUMAR, SAMANIEGO et ATTINGER, 2010 ;TOBIN et al., 2013 ;CHEN et al., 2017b Parmi les autres approches qui existent pour la modélisation du sol superficiel se trouvent certains concepts spécifiquement orientés vers l'hydrologie. C'est le cas par exemple de TOPMODEL, qui est spécifiquement construit pour modéliser le routage de l'eau dans le sol superficiel ainsi que sa propension à s'engorger, sur la base de la topographie (BEVEN et FREER, 2001 ;VINCENDON et al., 2010). ...
Les risques naturels en montagne font l'objet de mesures de prévention, souvent liées à des démarches de prévision. Dans certaines situations, la prévision de la survenue d'évènements liés à ces risques, voire la simple connaissance des processus physiques qui leur est associé, constitue un enjeu scientifique important compte-tenu de la grande complexité et de la forte hétérogénéité de ce milieu. La connaissance des mécanismes de formation des crues rapides sur les petits bassins versants englacés, ainsi que la perspective de leur prévision, est un exemple de ces risques difficilement maitrisables. La diversité des facteurs influençant les débits des rivières, leur complexité individuelle ainsi que celle de la manière dont ils interagissent, la forte variabilité spatio-temporelle des conditions météorologiques de la montagne ainsi que les modifications ayant lieu sur le long terme en raison du changement climatique font que ce phénomène nécessite une étude approfondie mobilisant des compétences pluri-disciplinaires, allant de la mesure de terrain au développement de modèles numériques prenant en compte les divers phénomènes liés à ce risque. Cette thèse s'inscrit dans le cadre d'un projet mis en oeuvre en partenariat avec les collectivités locales oevrant dans la vallée de Chamonix, voué à apporter un appui scientifique à la maitrise de ce risque. Au sein de ce projet, cette thèse porte sur le développement et le déploiement d'un modèle hydrologique prenant en compte la neige et les glaciers. Ce modèle se veut avoir deux objectifs : 1) servir d'outil de recherche permettant par exemple d'exploiter les mesures de terrain réalisées, en les confrontant aux résultats produits par ce modèle, et plus généralement de servir d'outil d'étude et de compréhension du fonctionnement de ce bassin, et 2) servir d'outil d'aide à la prévision des crues, en étant en mesure de fournir une prévision des débits de l'Arve à Chamonix à partir des données de prévision météorologique. L'exploitation des possibilités toujours grandissantes de la modélisation à bases physiques fait également partie des objectifs de cette thèse. En particulier, l'utilisation d'un modèle de neige à bilan d'énergie permettant notamment une représentation détaillée de l'interaction neige-glace a été mise en oeuvre, associée à l'exploitation des nombreuses mesures de terrain pour une évaluation en profondeur des résultats du modèle. Enfin, un déploiement expérimental de ce modèle en prévision a eu lieu à la fin de cette thèse.
... The model runs at hourly time step. The model has been cali- brated and evaluated in previous studies conducted in the same area showing very good capability to match streamflow observations of catchments of different sizes ( Kumar et al., 2010Kumar et al., , 2013Samaniego et al., 2010;Wöhling et al., 2013). This parameterization is also used for the present study (i.e., the same global calibrated parameters), but no additional tuning of the parameters has been performed. ...
The improvement of process representations in hydrological models is often only driven by the modelers' knowledge and data availability. We present a comprehensive comparison between two hydrological models of different complexity that is developed to support (1) the understanding of the differences between model structures and (2) the identification of the observations needed for model assessment and improvement. The comparison is conducted on both space and time and by aggregating the outputs at different spatiotemporal scales. In the present study, mHM, a process-based hydrological model, and ParFlow-CLM, an integrated subsurface-surface hydrological model, are used. The models are applied in a mesoscale catchment in Germany. Both models agree in the simulated river discharge at the outlet and the surface soil moisture dynamics, lending their supports for some model applications (drought monitoring). Different model sensitivities are, however, found when comparing evapotranspiration and soil moisture at different soil depths. The analysis supports the need of observations within the catchment for model assessment, but it indicates that different strategies should be considered for the different variables. Evapotranspiration measurements are needed at daily resolution across several locations, while highly resolved spatially distributed observations with lower temporal frequency are required for soil moisture. Finally, the results show the impact of the shallow groundwater system simulated by ParFlow-CLM and the need to account for the related soil moisture redistribution. Our comparison strategy can be applied to other models types and environmental conditions to strengthen the dialog between modelers and experimentalists for improving process representations in Earth system models.
... This study utilizes the mesoscale Hydrologic Model (mHM v5.8: Samaniego et al., 2017a), which is a grid-based spa-tially distributed hydrological model (Kumar et al., 2013(Kumar et al., , 2010Samaniego et al., 2010a, b). The model accounts for key hydrological processes such as canopy interception, soil moisture dynamics, surface and subsurface flow generation, snow melting, evapotranspiration and others. ...
The process of model evaluation is not only an integral part of
model development and calibration but also of paramount importance when
communicating modelling results to the scientific community and
stakeholders. The modelling community has a large and well-tested toolbox of
metrics to evaluate temporal model performance. In contrast, spatial
performance evaluation does not correspond to the grand availability of
spatial observations readily available and to the sophisticate model codes
simulating the spatial variability of complex hydrological processes. This
study makes a contribution towards advancing spatial-pattern-oriented model
calibration by rigorously testing a multiple-component performance metric.
The promoted SPAtial EFficiency (SPAEF) metric reflects three equally
weighted components: correlation, coefficient of variation and histogram
overlap. This multiple-component approach is found to be advantageous in
order to achieve the complex task of comparing spatial patterns. SPAEF, its
three components individually and two alternative spatial performance
metrics, i.e. connectivity analysis and fractions skill score, are applied
in a spatial-pattern-oriented model calibration of a catchment model in
Denmark. Results suggest the importance of multiple-component metrics
because stand-alone metrics tend to fail to provide holistic pattern
information. The three SPAEF components are found to be independent, which
allows them to complement each other in a meaningful way. In order to
optimally exploit spatial observations made available by remote sensing
platforms, this study suggests applying bias insensitive metrics which
further allow for a comparison of variables which are related but may differ in unit.
This study applies SPAEF in the hydrological context using the mesoscale
Hydrologic Model (mHM; version 5.8), but we see great potential across
disciplines related to spatially distributed earth system modelling.
... The applied upscaling rules are different for the various model parameters (e.g., the geometric mean for the porosity and soil hydraulic conductivity). Detailed information about model parameters and upscaling rules are provided in Samaniego et al. (2010), Kumar et al. (2010Kumar et al. ( , 2013a, and Zink et al. (2017). ...
Hydrologic models are usually calibrated using observed river runoff at catchment outlets. Streamflow, however, represents an integral response of the entire catchment and is observed at a few locations worldwide. Parameter estimation based on streamflow has the disadvantage that it does not consider the spatiotemporal variability of hydrologic states and fluxes such as evapotranspiration. Remotely sensed data, in contrast, include these variabilities and are broadly available. In this study, we assess the predictive skill of satellite‐derived land surface temperature (Ts) with respect to river runoff (Q). We developed a bias‐insensitive pattern‐matching criterion to focus the parameter optimization on spatial patterns of Ts. The proposed method is extensively tested in six distinct large German river basins and cross validated in 222 additional basins in Germany. We conclude that land surface temperature calibration outperforms random drawn parameter sets, which could be meaningful for calibrating hydrologic models in ungauged locations. A combined calibration with Q and Ts reduces the root mean squared error in the predicted evapotranspiration by 8% compared to flux tower observations but reduces the NSEs of the streamflow predictions by 6% on average for the six large basins. Our results show that patterns of Ts better constrain model parameters when considered in a calibration next to Q, which finally reduces parametric uncertainty. Parameter estimation framework using spatially distributed land surface temperature Assessment of performance of streamflow by calibration with land surface temperature only Improved parameter constraint with calibration using streamflow and land surface temperature
... Many error metrics have been used in the calibration of hydrologic models. For example, several studies [e.g., Kumar et al., 2010;Bock et al., 2016] used the averages of two NSE values computed with raw and logarithm daily streamflow series to account for errors in low flow as well as high flow. The use of a set of hydrologic signatures [Yilmaz et al., 2008] instead of streamflow time series potentially improves various aspects of the temporal streamflow pattern, e.g., low flow, high flow, and recession characteristics. ...
Estimating spatially distributed parameters remains one of the biggest challenges for large domain hydrologic modeling. Many large domain modeling efforts rely on spatially inconsistent parameter fields, e.g., patchwork patterns resulting from individual basin calibrations, parameter fields generated through default transfer functions that relate geophysical attributes to model parameters, or spatially constant, default parameter values. This paper provides an initial assessment of a multi-scale parameter regionalization (MPR) method over large geographical domains to derive seamless parameters in a spatially consistent manner. MPR applies transfer functions at the native scale of the geophysical data, and then scales these model parameters to the desired model resolution. We developed a stand-alone framework called MPR-flex for multi-model use and applied MPR-flex to the Variable Infiltration Capacity model to produce hydrologic simulations over the contiguous USA (CONUS). We first independently calibrate 531 basins across the CONUS to obtain a performance benchmark for each basin. To derive the CONUS parameter fields, we perform a joint MPR calibration using all but the poorest behaved basins to obtain a single set of transfer function parameters that are applied to the entire CONUS. Results show that the CONUS-wide calibration has similar performance compared to previous simulations using a patchwork quilt of partially calibrated parameter sets, but without the spatial discontinuities in parameters that characterize some previous CONUS-domain model simulations. Several avenues to improve CONUS-wide calibration remain, including selection of calibration basins, objective function formulation, as well as MPR-flex improvements including transfer function formations and scaling operator optimization.
... Conceptual models can be implemented as lumped or (semi-)distributed formulations (e.g. Kumar et al., 2010;Gao et al., 2014a;Fenicia et al., 2016). In spite of that they are sometimes collectively and inaccurately referred to as "lumped" models. ...
In hydrology, two somewhat competing philosophies form the basis of
most process-based models. At one endpoint of this continuum are detailed,
high-resolution descriptions of small-scale processes that are numerically
integrated to larger scales (e.g. catchments). At the other endpoint of the
continuum are spatially lumped representations of the system that express the
hydrological response via, in the extreme case, a single linear transfer
function. Many other models, developed starting from these two contrasting
endpoints, plot along this continuum with different degrees of spatial
resolutions and process complexities. A better understanding of the
respective basis as well as the respective shortcomings of different
modelling philosophies has the potential to improve our models. In this paper
we analyse several frequently communicated beliefs and assumptions to
identify, discuss and emphasize the functional similarity of the seemingly
competing modelling philosophies. We argue that deficiencies in model
applications largely do not depend on the modelling philosophy, although some
models may be more suitable for specific applications than others and vice
versa, but rather on the way a model is implemented. Based on the premises
that any model can be implemented at any desired degree of detail and that
any type of model remains to some degree conceptual, we argue that a
convergence of modelling strategies may hold some value for advancing the
development of hydrological models.
... Sin duda, los resultados obtenidos del modelado hidrológico dependen de diversos factores, como: (1) una adecuada representación espacial de los procesos hidrológicos, (2) el método utilizado en la parametrización del modelo, (3) el procedimiento para estimar de forma eficaz los parámetros del modelo, y (4) la calidad de la información Tecnología y Ciencias del Agua, vol. IV, núm. 5, noviembre-diciembre de 2013 Guerra-Cobián et al., Efecto de la discretización espacial sobre las simulaciones de caudal con el modelo distribuido CEQUEAU (Réméniéras, 1999;Singh y Woolhiser, 2002;Kumar et al., 2010;Guerra-Cobián et al., 2011). Diversos estudios de modelado hidrológico destacan que la respuesta de una cuenca es sensible a la heterogeneidad espacial de sus características físicas (topografía, textura de suelo, cobertura vegetal, etcétera), así como a la variabilidad espacio-temporal de los fenómenos meteorológicos (precipitación, temperatura, etcétera) (Krajewski et al., 1991;Koren et al., 1999;Grayson y Bloöschl, 2000;Bronstert et al., 2002). ...
This paper discusses the effect of the spatial discretization of a basin on flow simulations using the CEQUEAU model. This is a distributed model that was developed by the National Scientific Research Institute of the University of Quebec, Canada. CEQUEAU has been used in several countries to model runoff and by studies to compare models generated by the World Meteorological Organization (WMO). This model discretizes the basin into square elements, calculating water balance square-by-square, taking into account precipitation, temperature, percentage of lakes, percentage of forest, etc. The main objective of this work was to identify criteria to determine the optimal grid size for the discretization of a basin, assuming adequate representation of precipitation. Eight basins in the Mexican Republic with different sizes and different climates were analyzed. The physiographic characteristics were determined with SIG-Idrisi. In addition, CEQUEAU was calibrated and validated daily with various square sizes, evaluating the certainty level with numerical Nash criteria and graphs. Based on the results from the experimental design proposed, a suitable mathematical regression model was built to obtain the optimal spatial discretization grid size using easily-obtained physiographic parameters as explanatory variables.
... The KGE components also indicate a similar tendency. Such behavior is also observed in previous modeling studies [e.g.,Kumar et al., 2010;Lerat et al., 2012;Euser et al., 2013;Hrachowitz et al., 2014;Newman et al., 2015]. Additionally, a comparatively larger spread in these statistics is noticed when moving from calibration to evaluation period. ...
Increased availability and quality of near real-time observations provide the opportunity to improve understanding of predictive skills of hydrologic models. Recent studies have shown the limited capability of river discharge data alone to adequately constrain different components of distributed model parameterizations. In this study, the GRACE satellite-based total water storage (TWS) anomaly is used to complement the discharge data with the aim to improve the fidelity of mesoscale hydrologic model (mHM) through multivariate parameter estimation. The study is conducted on 83 European basins covering a wide range of hydroclimatic regimes. The model parameterization complemented with the TWS anomalies leads to statistically significant improvements in (1) discharge simulations during low-flow period, and (2) evapotranspiration estimates which are evaluated against independent data (FLUXNET). Overall, there is no significant deterioration in model performance for the discharge simulations when complemented by information from the TWS anomalies. However, considerable changes in the partitioning of precipitation into runoff components are noticed by in-/exclusion of TWS during the parameter estimation. Introducing monthly averaged TWS data only improves the dynamics of streamflow on monthly or longer time scales, which mostly addresses the dynamical behavior of the base flow reservoir. A cross-evaluation test carried out to assess the transferability of the calibrated parameters to other locations further confirms the benefit of complementary TWS data. In particular, the evapotranspiration estimates show more robust performance when TWS data are incorporated during the parameter estimation, in comparison with the benchmark model constrained against discharge only. This study highlights the value for incorporating multiple data sources during parameter estimation to improve the overall realism of hydrologic models and their applications over large domains.
... Watersheds are commonly spatially discretized in ecologic and hydrological studies. The purpose of spatial discretization is to objectively represent the differences in ecological or hydrological characteristics that exist within the watershed (Kumar et al., 2010;Hellebrand and van den Bos, 2008). By using spatial discretization, a watershed is divided into units, which are treated as statistical objects or calculated units for statistical analysis or simulation. ...
The spatial discretization of watersheds is an indispensable procedure for representing landscape variations in eco-hydrological research, representing the contrast between reality and data-supported models. When discretizing a watershed, it is important to construct a scheme of a moderate number of discretized factors while adequately considering the actual eco-hydrological processes, especially in regions with unique eco-hydrological features and intense human activities. Because of their special lithological and pedologic characteristics and widespread man-made vegetation, discretization of watersheds in the Loess Plateau in Northern China is a challenge. In order to simulate the rainfall-runoff process, a watershed in the Loess Plateau, referred as Ansai, was spatially discretized into new units called land type units. These land type units were delineated under a scheme of factors including land use, vegetation condition, soil type and slope. Instead of using units delineated by overlaying land use and soil maps, the land type units were used in the Soil and Water Assessment Tool (SWAT). Curve numbers were assigned and adjusted to simulate runoff, using the US Natural Resources Conservation Service (NRCS) curve number method. The results of the runoff simulation better matched actual observations. Compared to the results that used the original units, the coefficient of determination ( R <sup>2</sup>) and the Nash-Sutcliffe coefficient ( E <sub>NS</sub>) for monthly flow simulation increased from 0.655–0.713 and 0.271–0.550 to 0.733–0.745 and 0.649–0.703, respectively. This method of delineating into land type units is an easy operation and suitable approach for eco-hydrological studies in the Chinese Loess Plateau and other similar regions. It can be further applied in soil erosion simulation and the eco-hydrological assessment of re-vegetation.
... Esta técnica reduz substancialmente o número de parâmetros do modelo a serem calibrados, uma vez que um conjunto de parâmetros precisa ser estimado para algumas poucas URHs, ao invés de ser estimado para cada célula ou minibacia que compõe o modelo. A aplicação deste tipo de técnica pode ser encontrada em muitos estudos (por exemplo: Beldring et al., 2003;Das et al., 2008;Kumar et al., 2010) e tem sido aplicada em modelos como SWAT (Arnold et al., 1998) e MGB-IPH (Fan e Collischonn, 2014). ...
A representação da variação espacial de características físicas das bacias hidrográficas é um dos grandes desafios da aplicação de modelos hidrológicos distribuídos. Neste sentido, uma técnica muito empregada é a definição de Unidades de Resposta Hidrológica (URHs), que são regiões hidrologicamente homogêneas dentro da bacia. Um mapa de URHs é, geralmente, definido a partir da combinação de mapas de tipo e de uso do solo, podendo ainda incluir outras informações como geologia e topografia. Este trabalho apresenta a derivação de um mapa de URHs para toda a América do Sul, a partir de informações de uso e tipo de solo em diferentes escalas, permitindo a sua aplicação direta em modelagem hidrológica de grandes bacias hidrográficas. Além de ser um produto específico para atender necessidades hidrológicas em modelos, ele reduz o tempo de preparação dos dados de entrada. O mapa final tem sido utilizado em diversas aplicações com o modelo MGB-IPH. Neste trabalho ele é testado na simulação da bacia hidrográfica do rio Paraná. Os resultados da simulação nesta bacia com uso do mapa foram satisfatórios, e pretende-se usar o mapa em diversas aplicações futuras.
... Numerous water resource indicators (WRIs) were proposed to depict flow components, such as average monthly, seasonal and annual flows, magnitude and timing of peak or low flows (Shrestha et al., 2013). Moreover, the prediction of extreme events (floods and droughts) was taken more and more seriously because of their disastrous damages to society, economy and environment, especially in the arid and semi-arid regions (Smakhtin, 2001;Coulibaly et al., 2001;Held et al., 2005;Kumar et al., 2010). However, the simulation performances of WRIs were still far from satisfactory, particularly for the low flow events (Wenger et al., 2010;Staudinger et al., 2011;Pushpalatha et al., 2012;Shrestha et al., 2013). ...
Flow regimes (e.g., magnitude, frequency, variation, duration, timing and rating of change) play a critical role in water supply and flood control, environmental processes, as well as biodiversity and life history patterns in the aquatic ecosystem. The traditional flow magnitude-oriented calibration of hydrological model was usually inadequate to well capture all the characteristics of observed flow regimes. In this study, we simulated multiple flow regime metrics simultaneously by coupling a distributed hydrological model with an equally weighted multi-objective optimization algorithm. Two headwater watersheds in the arid Hexi Corridor were selected for the case study. Sixteen metrics were selected as optimization objectives, which could represent the major characteristics of flow regimes. Model performance was compared with that of the single objective calibration. Results showed that most metrics were better simulated by the multi-objective approach than those of the single objective calibration, especially the low and high flow magnitudes, frequency and variation, duration, maximum flow timing and rating. However, the model performance of middle flow magnitude was not significantly improved because this metric was usually well captured by single objective calibration. The timing of minimum flow was poorly predicted by both the multi-metric and single calibrations due to the uncertainties in model structure and input data. The sensitive parameter values of the hydrological model changed remarkably and the simulated hydrological processes by the multi-metric calibration became more reliable, because more flow characteristics were considered. The study is expected to provide more detailed flow information by hydrological simulation for the integrated water resources management, and to improve the simulation performances of overall flow regimes.
Transit time‐based water quality models using StorAge Selection (SAS) functions are crucial for nitrate (NO 3 ⁻ ) management. However, relying solely on instream NO 3 ⁻ concentration for model calibration can result in poor parameter identifiability. This is due to the interaction, or correlation, between transport parameters, such as SAS function parameters, and denitrification rate, which challenges accurate parameters identification and description of catchment‐scale hydrological processes. To tackle this issue, we conducted three Monte‐Carlo experiments for a German mesoscale catchment by calibrating a SAS‐based model with daily instream NO 3 ⁻ concentrations (Experiment 1), monthly instream stable water isotopes (e.g. δ ¹⁸ O) (Experiment 2) and both datasets (Experiment 3). Our findings revealed comparable ranges of SAS transport parameters and median water transit times (TT 50 ) across the experiments. This suggests that, despite their distinct reactive or conservative nature, and sampling strategies, the NO 3 ⁻ and δ ¹⁸ O time series offer similar information for calibration. However, the absolute values of transport parameters and TT 50 time series, as well as the degree of parameter interaction differed. Experiment 1 showed greater interaction between certain transport parameters and denitrification rate, leading to greater equifinality. Conversely, Experiment 3 yielded reduced parameters interaction, which enhanced transport parameters identifiability and decreased uncertainty in TT 50 time series. Hence, even a modest effort to incorporate only monthly δ ¹⁸ O values in model calibration for highly frequent NO 3 ⁻ , improved the description of hydrological transport. This study showcased the value of combining NO 3 ⁻ and δ ¹⁸ O model results to improve transport parameter identifiability and model robustness, which ultimately enhances NO 3 ⁻ management strategies.
Transit time distributions (TTDs) of streamflow are useful descriptors for understanding flow and solute transport in catchments. Catchment-scale TTDs can be modeled using tracer data (e.g. oxygen isotopes, such as δ18O) in inflow and outflows by employing StorAge Selection (SAS) functions. However, tracer data are often sparse in space and time, so they need to be interpolated to increase their spatiotemporal resolution. Moreover, SAS functions can be parameterized with different forms, but there is no general agreement on which one should be used. Both of these aspects induce uncertainty in the simulated TTDs, and the individual uncertainty sources as well as their combined effect have not been fully investigated. This study provides a comprehensive analysis of the TTD uncertainty resulting from 12 model setups obtained by combining different interpolation schemes for δ18O in precipitation and distinct SAS functions. For each model setup, we found behavioral solutions with satisfactory model performance for in-stream δ18O (KGE > 0.55, where KGE refers to the Kling–Gupta efficiency). Differences in KGE values were statistically significant, thereby showing the relevance of the chosen setup for simulating TTDs. We found a large uncertainty in the simulated TTDs, represented by a large range of variability in the 95 % confidence interval of the median transit time, varying at the most by between 259 and 1009 d across all tested setups. Uncertainty in TTDs was mainly associated with the temporal interpolation of δ18O in precipitation, the choice between time-variant and time-invariant SAS functions, flow conditions, and the use of nonspatially interpolated δ18O in precipitation. We discuss the implications of these results for the SAS framework, uncertainty characterization in TTD-based models, and the influence of the uncertainty for water quality and quantity studies.
Like many other regions in central Europe, Germany experienced sequential summer droughts from 2015-2018. As one of the environmental consequences, river nitrate concentrations have exhibited significant changes in many catchments. However, catchment nitrate responses to the changing weather conditions have not yet been mechanistically explored. Thus, a fully distributed, process-based catchment Nitrate model (mHM-Nitrate) was used to reveal the causal relations in the Bode catchment, of which river nitrate concentrations have experienced contrasting trends from upstream to downstream reaches. The model was evaluated using data from six gauging stations, reflecting different levels of runoff components and their associated nitrate-mixing from upstream to downstream. Results indicated that the mHM-Nitrate model reproduced dynamics of daily discharge and nitrate concentration well, with Nash-Sutcliffe Efficiency ≥ 0.73 for discharge and Kling-Gupta Efficiency ≥ 0.50 for nitrate concentration at most stations. Particularly, the spatially contrasting trends of nitrate concentration were successfully captured by the model. The decrease of nitrate concentration in the lowland area in drought years (2015-2018) was presumably due to (1) limited terrestrial export loading (ca. 40% lower than that of normal years 2004-2014), and (2) increased in-stream retention efficiency (20% higher in summer within the whole river network). From a mechanistic modelling perspective, this study provided insights into spatially heterogeneous flow and nitrate dynamics and effects of sequential droughts, which shed light on water-quality responses to future climate change, as droughts are projected to be more frequent.
Spatial parameter fields are required to model hydrological processes across diverse landscapes. Transfer functions are often used to relate parameters to spatial catchment attributes, introducing large uncertainties. Quantifying these uncertainties remains a key challenge for large‐scale modeling. This paper extends the multiscale parameter regionalization (MPR) technique to consider parameter uncertainties. We evaluate this method of producing nationally consistent parameter fields, which maintain a constant relationship between model parameters and catchment attributes, across 437 catchments in Great Britain (GB). By sampling multiple transfer function parameters, we produce thousands of possible model parameter fields which are constrained within an uncertainty framework. This is compared to spatially homogeneous parameter sets constrained for individual catchments. The nationally consistent MPR parameter fields perform well (KGE* > 0.75) across 60% of catchments. Performance is similar or better than catchment‐constrained parameters (KGE* drop < 0.1) across 82% of catchments. Advantages of our national parameter fields include (a) improved representation of flows within catchments, (b) more robust performance between calibration and evaluation periods, and (c) spatial parameter fields reflecting hydrologically meaningful variation in catchment characteristics. By including uncertainties, we show that hydrographs produced using MPR have smaller uncertainty bounds which are better able to encompass flows. As the first application of MPR to both the DECIPHeR modeling framework and GB, we developed transfer functions and identified key catchment attributes to constrain model parameters, which are transferrable to other models alongside the addition of uncertainty. Methodologies presented here are informative for future regionalization efforts in GB and elsewhere.
Although models are one of the most powerful tools for watershed management, their effectiveness is limited by prediction uncertainties resulting from not only model input data but also spatial discretization. In this paper, Hydrological Simulation Program-Fortran (HSPF) models were constructed for the Linyi watershed according to three segmentation approaches including model segments based on differences in: (1) sub-watershed, (2) meteorological station, and (3) physical characteristics. Then the static sensitivity method and dynamic sensitivity method were employed to evaluate the effect of the segmentation approach on model performance and parameters of HSPF. The main conclusions were: (1) modeling with 12 segments had the best simulation efficiency and the corresponding estimated parameters had a certain representation within the Linyi watershed; (2) HSPF model performance was significantly affected by the segmentation approach, especially by the model segmentation construction process which considering a meteorological station or not; (3) parameters INTFW (interflow inflow parameter), lower zone nominal storage, and upper zone nominal storage (UZSN) were most affected by the model segmentation approach, while parameter AGWRC (groundwater recession coefficient) changed indistinctly; (4) parameters UZSN and INTFW had the same variation tendency whenever the segmentation approach changed.
Understanding of rainfall–runoff model performance under non-stationary hydroclimatic conditions is limited. This study compared lumped (IHACRES), semi-distributed (HEC-HMS) and fully-distributed (SWATgrid) hydrological models to determine which most realistically simulates runoff in catchments where non-stationarity in rainfall–runoff relationships exists. The models were calibrated and validated under different hydroclimatic conditions (Average, Wet and Dry) for two heterogeneous catchments in southeast Australia (SEA). SWATgrid realistically simulates runoff in the smaller catchment under most hydroclimatic conditions but fails when the model is calibrated in Dry conditions and validated in Wet. All three models perform poorly in the larger catchment irrespective of hydroclimatic conditions. This highlights the need for more research aimed at improving the ability of hydrological models to realistically incorporate the physical processes causing non-stationarity in rainfall–runoff relationships. Although the study is focussed on SEA, the insights gained are useful for all regions which experience large hydroclimatic variability and multi-year/decadal droughts.
Probabilistic predictions from hydrological models, including rainfall-runoff models, provide valuable information for water and environmental resource risk management. However, traditional “deterministic” usage of rainfall-runoff models remains prevalent in practical applications, in many cases because probabilistic predictions are perceived to be difficult to generate. This paper introduces a simplified approach for hydrological model inference and prediction that bridges the practical gap between “deterministic” and “probabilistic” techniques. This approach combines the Least Squares (LS) technique for calibrating hydrological model parameters with a simple method-of-moments (MoM) estimator of error model parameters (here, the variance and lag-1 autocorrelation of residual errors). A case study using two conceptual hydrological models shows that the LS-MoM approach achieves probabilistic predictions with similar predictive performance to classical maximum-likelihood and Bayesian approaches—but is simpler to implement using common hydrological software and has a lower computational cost. A public web-app to help users implement the simplified approach is available.
Diagnostics of hydrological models are pivotal for a better understanding of catchment functioning, and the analysis of dominating model parameters plays a key role for region-specific calibration or parameter transfer. A major challenge in the analysis of parameter sensitivity is the assessment of both temporal and spatial differences of parameter influences on simulated streamflow response. We present a methodological approach for global sensitivity analysis of hydrological models. The multilevel approach is geared towards complementary forms of streamflow response targets, and combines sensitivity analysis directed to hydrological fingerprints, i.e. temporally independent and temporally aggregated characteristics of streamflow (INDPAS), with the conventional analysis of the temporal dynamics of parameter sensitivity (TEDPAS).
The approach was tested in 14 mesoscale headwater catchments of the Ruhr River in western Germany using simulations with the spatially distributed hydrological model mHM. The multilevel analysis with diverse response characteristics allowed us to pinpoint parameter sensitivity patterns much more clearly as compared to using TEDPAS alone. It was not only possible to identify two dominating parameters, for soil moisture dynamics and evapotranspiration, but we could also disentangle the role of these and other parameters with reference to different streamflow characteristics. The combination of TEDPAS and INDPAS further allowed us to detect regional differences in parameter sensitivity and in simulated hydrological functioning, despite the rather small differences in the hydroclimatic and topographic setting of the Ruhr headwaters.
Long-term, high-resolution data about hydrologic fluxes and states are needed
for many hydrological applications. Because continuous large-scale
observations of such variables are not feasible, hydrologic or land surface
models are applied to derive them. This study aims to analyze and provide a
consistent high-resolution dataset of land surface variables over Germany,
accounting for uncertainties caused by equifinal model parameters. The
mesoscale Hydrological Model (mHM)
is employed to derive an ensemble (100
members) of evapotranspiration, groundwater recharge, soil moisture, and runoff
generated at high spatial and temporal resolutions (4 km and daily,
respectively) for the period 1951–2010. The model is cross-evaluated against
the observed daily streamflow in 222 basins, which are not used for model
calibration. The mean (standard deviation) of the ensemble median
Nash–Sutcliffe efficiency estimated for these basins is 0.68 (0.09) for daily
streamflow simulations. The modeled evapotranspiration and soil moisture
reasonably represent the observations from eddy covariance stations. Our
analysis indicates the lowest parametric uncertainty for evapotranspiration,
and the largest is observed for groundwater recharge. The uncertainty of the
hydrologic variables varies over the course of a year, with the exception of
evapotranspiration, which remains almost constant. This study emphasizes the
role of accounting for the parametric uncertainty in model-derived
hydrological datasets.
Travel-time distributions are a comprehensive tool for the characterization
of hydrological system dynamics. Unlike the streamflow hydrograph, they
describe the movement and storage of water within and throughout the
hydrological system. Until recently, studies using such travel-time
distributions have generally either been applied to lumped models or to
real-world catchments using available time series, e.g., stable isotopes.
Whereas the former are limited in their realism and lack information on the
spatial arrangements of the relevant quantities, the latter are limited in
their use of available data sets. In our study, we employ the spatially
distributed mesoscale Hydrological Model (mHM) and apply it to a catchment in
central Germany. Being able to draw on multiple large data sets for
calibration and verification, we generate a large array of spatially
distributed states and fluxes. These hydrological outputs are then used to
compute the travel-time distributions for every grid cell in the modeling
domain. A statistical analysis indicates the general soundness of the
upscaling scheme employed in mHM and reveals precipitation, saturated soil
moisture and potential evapotranspiration as important predictors for
explaining the spatial heterogeneity of mean travel times. In addition, we
demonstrate and discuss the high information content of mean travel times for
characterization of internal hydrological processes.
Der Austausch von Wissen und Information zwischen verschiedenen gesellschaftlichen Gruppen ist oft nicht trivial. Vertreter aus der Öffentlichkeit, verschiedenen Fachkreisen und Behörden oder aus der Wissenschaft generieren sehr unterschiedliches Wissen unter Einbeziehung von unterschiedlichen Graden der Problemorientierung und in ihrer jeweiligen Sprache. Zur Überwindung dieser Barrieren stehen verschiedene Instrumente zur Verfügung. In diesem Artikel werden drei weitverbreitete Formen des Wissenstransfers diskutiert: (1) Assessments mit ihren verschiedenen Formen z. B. auf unterschiedlichen räumlichen Skalen, (2) Indikatoren mit möglichen Rahmenkonzepten, Indikatorensätze und Formen der Evaluierung und (3) web-basierte Plattformen als einfache Möglichkeit der Verbreitung von aktuellen Informationen. Dabei werde zwei Beispiele ausführlich dargestellt, nämlich das am Klimabüro für Polargebiete und Meeresspielgel konzipierte Meereisportal und der am Mitteldeutschen Klimabüro entwickelte Deutsche Dürremonitor.
Environmental models have become increasingly complex with greater attention being given to the spatially distributed representation of processes. Distributed models have large numbers of parameters to be specified, which is typically done either by recourse to a priori methods based on observable physical watershed characteristics, by calibration to watershed input-state-output data, or by some combination of both. In the case of calibration, the high dimensionality of the parameter search space poses a significant identifiability problem. This article discusses how this problem can be addressed, utilizing additional information about the parameters through a process known as regularization. Regularization, in its broadest sense, is a mathematical technique that utilizes additional information or constraints about the parameters to reduce problems related to over-parameterization. This article develops and applies a regularization approach to the calibration of a version of the Hydrology Laboratory Distributed Hydrologic Model (HL-DHM) developed by the US National Weather Service. A priori parameter estimates derived using the approach by Koren et al. (2000) were used to develop regularization relationships to constrain the feasible parameter space and enable existing global optimization techniques to be applied to solve the calibration problem. In a case study for the Blue River basin, the number of unknowns to be estimated was reduced from 858 to 33, and this calibration strategy improved the model performance while preserving the physical realism of the model parameters. Our results also suggest that the commonly used parameter field ``multiplier'' approach may often not be appropriate.
A Monte Carlo study of a physically based distributed-parameter hydrologic model is described. The catchment model simulates overland flow and streamflow, and it is based on the kinematic wave concept. Soil Conservation Service curves are used to model rainfall excess within the basin. The model was applied to the Ralston Creek watershed, a small (7.5 km2) rural catchment in eastern Iowa. Sensitivity of the model response with respect to rainfall-input spatial and temporal sampling density was investigated. The input data were generated by a space-time stochastic model of rainfall. The generated rainfall fields were sampled by the varied-density synthetic rain gauge networks. The basin response, based on 5-min increment input data from a network of high density with about 1 gauge per 0.1 km2, was assumed to be the ``ground truth,'' and other results were compared against it. Included in the study was also a simple lumped parameter model based on the unit hydrograph concept. Results were interpreted in terms of hydrograph characteristics such as peak magnitude, time-to-peak, and total runoff volume. The results indicate higher sensitivity of basin response with respect to the temporal resolution than to the spatial resolution of the rainfall data. Also, the frequency analysis of the flood peaks shows severe underestimation by the lumped model. This may have implications for the design of hydraulic structures.
The problem of selecting appropriate objective functions for the identifica- tion of a lumped conceptual rainfall-runoff model is investigated, focusing on the value of the model in an operational setting. A probability-distributed soil moisture model is coupled with a linear parallel routing scheme, and conditioned on rainfall- runoff observations from three catchments in the southeast of England. Using an abstraction control problem, which requires accurate simulation of the intermediate flow range, it is shown that using the traditional RMSE fit criterion, produces operationally sub-optimal predictions. This is true in the identification period, when applied to a testing period, and to proxy catchment data. Using a second case study of the Leaf River in Mississippi (USA), where the focus changes to predicting flood peaks over a specified threshold, also suggests that the relevant flood threshold should govern the objective function choice. It is concluded that, due to limitations in the structure of the employed model, it would be counter-productive to try to achieve a good all-round representation of the rainfall-runoff processes, and that a more em- pirical approach to identification may be preferred for specific forecasting problems. This leaves us with the question of how far hydrological realism should be sacrificed in favour of purpose-driven objective functions.
This paper is aimed at improving the efficiency of model uncertainty
analyses that are conditioned on measured calibration data.
Specifically, the focus is on developing an alternative methodology to
the generalized likelihood uncertainty estimation (GLUE) technique when
pseudolikelihood functions are utilized instead of a traditional
statistical likelihood function. We demonstrate for multiple calibration
case studies that the most common sampling approach utilized in GLUE
applications, uniform random sampling, is much too inefficient and can
generate misleading estimates of prediction uncertainty. We present how
the new dynamically dimensioned search (DDS) optimization algorithm can
be used to independently identify multiple acceptable or behavioral
model parameter sets in two ways. DDS could replace random sampling in
typical applications of GLUE. More importantly, we present a new,
practical, and efficient uncertainty analysis methodology called
DDS-approximation of uncertainty (DDS-AU) that quantifies prediction
uncertainty using prediction bounds rather than prediction limits.
Results for 13, 14, 26, and 30 parameter calibration problems show that
DDS-AU can be hundreds or thousands of times more efficient at finding
behavioral parameter sets than GLUE with random sampling. Results for
one example show that for the same limited computational effort, DDS-AU
prediction bounds can simultaneously be smaller and contain more of the
measured data in comparison to GLUE prediction bounds. We also argue and
then demonstrate that within the GLUE framework, when behavioral
parameter sets are not sampled frequently enough, Latin hypercube
sampling does not offer any improvements over simple random sampling.
Deterministic hydrological models are limited in their ability to model low-flows, but as they are increasingly being used for low-flow studies, a review is timely if they are to be used for this purpose. The representations of the physical processes that govern low-flows are described for a selection of models developed and/or used in Canada. The models vary in the extent to which they incorporate low-flow processes, such as drawdown of storage in lakes, stream channels and wetlands, riparian evapotranspiration, freeze-up and bank storage. The representation of groundwater in the models is not physically-based and the hydrologic models must be coupled with distributed groundwater models to answer questions about groundwater withdrawals. Most models have not been rigorously tested for low-flow simulations. Considerable work remains to be done, especially in the adequate representation of low-flow processes, and in evaluation of models for low-flow studies by using low-flow specific evaluation criteria.
The present paper provides a brief review of statistical models that are commonly used in the estimation of low flows both at sites with a reliable streamflow record and sites remote from data sources. Opportunities are identified for the regional estimation of low-flow characteristics at ungauged sites. The adaptation of the neighbourhood regionalization approach, commonly used in regional flood frequency analysis, can be extended to low-flow variables. Estimation approaches extending the usefulness of recession information in regional low-flow frequency analysis to ungauged sites using a canonical correlation analysis approach for the identification of hydrological neighbourhoods is described. The validity of recession parameters when estimated from very short hydrological data records is also discussed. Promising new directions for future research in the field of statistical low-flow frequency estimation are identified.
The majority of pedotransfer function (PTFs) published for estimating water retention characteristics (WRC) use data on soil texture, bulk density, and organic matter content (OM) as predictors. For soil hydrological modeling on a regional scale, in particular the derivation of appropriate values for a PTF parameterization can be difficult where organic C data are missing. Assuming the indirect interdependency between OM and bulk density, a new PTF has been developed that estimates the WRC using only soil texture and bulk density data. To achieve a regression-based reproduction of the correlations, a calibration was chosen that connects the parameters of the van Genuchten equation with the data on bulk density and soil texture, using linear and nonlinear relationships. More than 90% of the variability in measured soil water contents was explained by the new model. The validity of the PTF was tested with a data set of 147 measured WRCs (r(2) = 0.94). Compared with another frequently used PTF model, which uses the organic C content as an additional predictor, the new model provided comparable or slightly better predictions of the WRCs.
The problem of selecting appropriate objective functions for the identification of a lumped conceptual rainfall–runoff model is investigated, focusing on the value of the model in an operational setting. A probability-distributed soil moisture model is coupled with a linear parallel routing scheme, and conditioned on rainfall–runoff observations from three catchments in the southeast of England. Using an abstraction control problem, which requires accurate simulation of the intermediate flow range, it is shown that using the traditional RMSE fit criterion, produces operationally sub-optimal predictions. This is true in the identification period, when applied to a testing period, and to proxy catchment data. Using a second case study of the Leaf River in Mississippi (USA), where the focus changes to predicting flood peaks over a specified threshold, also suggests that the relevant flood threshold should govern the objective function choice. It is concluded that, due to limitations in the structure of the employed model, it would be counter-productive to try to achieve a good all-round representation of the rainfall–runoff processes, and that a more empirical approach to identification may be preferred for specific forecasting problems. This leaves us with the question of how far hydrological realism should be sacrificed in favour of purpose-driven objective functions.
There are many procedures in the available literature to perform
prediction in ungauged basins. Commonly, the Euclidean metric is used as
a proxy of the hydrologic dissimilarity. Here we propose a procedure to
find a metric on the basis of dissimilarity measures that are estimated
from pairwise empirical copula densities of runoff. A metric is then
defined in an transformed space of basin descriptors, whose
parameterization is obtained with a variance reducing technique. A
hydrologic model was run in an ungauged basin with sets of global
parameters obtained from the k nearest neighboring donor basins using
various metrics to take into account the uncertainty of its
parameterization. Hydrologic model parameters were regionalized with a
multiscale parameter regionalization technique whose transfer function
parameters were found via calibration. The streamflow in an ungauged
basin was found as an ensemble streamflow prediction to account for the
uncertainties of the transfer function parameters as well as those to
define the metric. This technique was applied in 38 German basins
ranging in size from 70 to 4000 km2. For each basin, a number
of catchment descriptors and several climatic indices were quantified,
e.g., mean slope, aspect, shape factor, mean elevation, and mean monthly
temperature in January, among others. Daily streamflow time series
correspond to the period from 1961 to 2000. Simulated daily discharge
was validated with a Jackknife cross-validation technique.
Nash-Sutcliffe efficiencies obtained in this way ranged between 0.76 and
0.86. These results suggested that the proposed technique would produce
reasonable results in ungauged basins.
The requirements for hydrological models have increased considerably during the previous decades to cope with the resolution of extensive remotely sensed data sets and a number of demanding applications. Existing models exhibit deficiencies such as overparameterization, the lack of an effective technique to integrate the spatial heterogeneity of physiographic characteristics, and the nontransferability of parameters across scales and locations. A multiscale parameter regionalization (MPR) technique is proposed as a way to address these issues simultaneously. Using this technique, parameters at a coarser scale, in which the dominant hydrological processes are represented, are linked with their corresponding ones at a finer resolution in which input data sets are available. The linkage is done with upscaling operators such as the harmonic mean, among others. Parameters at the finer scale are regionalized through nonlinear transfer functions which link basin predictors with global parameters to be determined through calibration. MPR was compared with a standard regionalization (SR) method in which basin predictors instead of model parameters are first aggregated. Both methods were tested in a basin located in Germany using a distributed hydrologic model. Results indicate that MPR is superior to SR in many respects, especially if global parameters are transferred from coarser to finer scales. Furthermore, MPR, as opposed to SR, preserves the spatial variability of state variables and conserves the mass balance with respect to a control scale. Cross-validation tests indicate that the transferability of the global parameters to ungauged locations is possible.
1] A new global optimization algorithm, dynamically dimensioned search (DDS), is introduced for automatic calibration of watershed simulation models. DDS is designed for calibration problems with many parameters, requires no algorithm parameter tuning, and automatically scales the search to find good solutions within the maximum number of user-specified function (or model) evaluations. As a result, DDS is ideally suited for computationally expensive optimization problems such as distributed watershed model calibration. DDS performance is compared to the shuffled complex evolution (SCE) algorithm for multiple optimization test functions as well as real and synthetic SWAT2000 model automatic calibration formulations. Algorithms are compared for optimization problems ranging from 6 to 30 dimensions, and each problem is solved in 1000 to 10,000 total function evaluations per optimization trial. Results are presented so that future modelers can assess algorithm performance at a computational scale relevant to their modeling case study. In all four of the computationally expensive real SWAT2000 calibration formulations considered here (14, 14, 26, and 30 calibration parameters), results show DDS to be more efficient and effective than SCE. In two cases, DDS requires only 15–20% of the number of model evaluations used by SCE in order to find equally good values of the objective function. Overall, the results also show that DDS rapidly converges to good calibration solutions and easily avoids poor local optima. The simplicity of the DDS algorithm allows for easy recoding and subsequent adoption into any watershed modeling application framework. Citation: Tolson, B. A., and C. A. Shoemaker (2007), Dynamically dimensioned search algorithm for computationally efficient watershed model calibration, Water Resour. Res., 43, W01413, doi:10.1029/2005WR004723.
Several contributions to the hydrological literature have brought into question the continued usefulness of the classical paradigm for hydrologic model calibration. With the growing popularity of sophisticated "physically based" watershed models (e.g., land-surface hydrology and hydrochemical models) the complexity of the calibration problem has been multiplied many fold. We disagree with the seemingly widespread conviction that the model calibration problem will simply disappear with the availability of more and better field measurements. This paper suggests that the emergence of a new and more powerful model calibration paradigm must include recognition of the inherent multiobjective nature of the problem and must explicitly recognize the role of model error. The results of our preliminary studies are presented. Through an illustrative case study we show that the multiobjective approach is not only practical and relatively simple to implement but can also provide useful information about the limitations of a model.
River stage or flow rates are required for the design and evaluation of hydraulic structures. Most river reaches are ungauged and a methodology is needed to estimate the rates of flow, at specific locations in streams where no measurements are available.
Flood-routing techniques are utilised to estimate the stages, or rates of flow, in order to predict flood wave propagation along river reaches. Models can be developed for gauged catchments and their parameters related to physical characteristics such as slope, reach width, reach length so that the approach can be applied to ungauged catchments within the same region.
The objective of this study is to assess the Muskingum-Cunge method for flow routing in ungauged river reaches, both with and without lateral inflows. The Muskingum-Cunge method was assessed using catchment-derived parameters for use in ungauged river reaches. Three sub-catchments in the Thukela catchment in KwaZulu-Natal, South Africa were selected for analyses, with river lengths of 4, 21 and 54 km. The slopes of the river reaches and reach lengths were derived from a digital elevation model. Manning\'s roughness coefficients were estimated from field observations. Flow variables such as velocity, hydraulic radius, wetted perimeters and flow depth were determined both from empirical equations and assumed cross-sections
of the reaches. Lateral inflows to long river reaches were estimated from the Saint-Venant equation. The performance of the methods was evaluated by comparing both graphically and statistically the simulated and observed hydrographs. The results obtained show that the computed outflow hydrographs generated using the Muskingum-Cunge method, with variables estimated using both the empirical relationships or assumed cross-sectional shapes, resulted in reasonably accurate computed outflow hydrographs with respect to volume, peak discharge, timing of peak flow and shape of the hydrograph.
From this study, it is concluded that the Muskingum-Cunge method can be applied to route floods in ungauged catchments using derived variables in the Thukela catchment and it is postulated that the method can be used to route floods in other ungauged rivers in South Africa. Water SA Vol.32 (3) 2006: pp.379-388
Most studies on the impact of climate change on regional water resources focus on long-term average flows or mean water availability,
and they rarely take the effects of altered human water use into account. When analyzing extreme events such as floods and
droughts, the assessments are typically confined to smaller areas and case studies. At the same time it is acknowledged that
climate change may severely alter the risk of hydrological extremes over large regional scales, and that human water use will
put additional pressure on future water resources. In an attempt to bridge these various aspects, this paper presents a first-time
continental, integrated analysis of possible impacts of global change (here defined as climate and water use change) on future
flood and drought frequencies for the selected study area of Europe. The global integrated water model WaterGAP is evaluated
regarding its capability to simulate high and low-flow regimes and is then applied to calculate relative changes in flood
and drought frequencies. The results indicate large ‘critical regions’ for which significant changes in flood or drought risks
are expected under the proposed global change scenarios. The regions most prone to a rise in flood frequencies are northern
to northeastern Europe, while southern and southeastern Europe show significant increases in drought frequencies. In the critical
regions, events with an intensity of today's 100-year floods and droughts may recur every 10–50 years by the 2070s. Though
interim and preliminary, and despite the inherent uncertainties in the presented approach, the results underpin the importance
of developing mitigation and adaptation strategies for global change impacts on a continental scale.
For regional aquifer modeling it is often necessary to produce maps of the distribution of the transmissivity in the aquifer, for example, as initial input for the calibration phase of the model, either by automatic or by trial and error procedures. Such estimations must be based on all possible information available in the field; in many instances, direct transmissivity measurements from pumping tests are scarce, whereas indirect estimations based on specific capacity data are more numerous. It is, however, possible to use jointly both types of data when a geostatistical estimation technique is used. Four such methods will be compared here: (1) kriging combined with linear regression, (2) cokriging, (3) kriging with an external drift, and (4) kriging with a guess field. This comparison is made both on a set of real field data and on a theoretical case, where the “true” solution is known.
The impact of climate change on flooding in the river Meuse is assessed on a daily basis using spatially and temporally changed climate patterns and a hydrological model with three different spatial resolutions. This is achieved by selecting a hydrological modelling framework and implementing appropriate model components, derived in an earlier study, into the selected framework (HBV). Additionally, two other spatial resolutions for the hydrological model are used to evaluate the sensitivity of the model results to spatial model resolution and to allow for a test of the model appropriateness procedure. Generations of a stochastic precipitation model under current and changed climate conditions have been used to assess the climate change impacts. The average and extreme discharge behaviour at the basin outlet is well reproduced by the three versions of the hydrological model in the calibration and validation, the results become somewhat better with increasing model resolution. The model results with synthetic precipitation under current climate conditions show a small overestimation of average discharge behaviour and a considerable underestimation of extreme discharge behaviour. The underestimation of extreme discharges is caused by the small-scale character of the observed precipitation input at the sub-basin scale. The general trend with climate change is a small decrease of the average discharge and a small increase of discharge variability and extreme discharges. The variability in extreme discharges for climate change conditions increases with respect to the simulations for current climate conditions. This variability results both from the stochasticity of the precipitation process and the differences between the climate models. The total uncertainty in river flooding with climate change (over 40%) is much larger than the change with respect to current climate conditions (less than 10%). However, climate changes are systematic changes rather than random changes and thus the large uncertainty range will be shifted to another level corresponding to the changed average situation.
Predictions of probabilities and magnitudes of extreme events are essential for water management. One approach for flood estimation is the use of conceptual runoff models. This approach, however, can be questioned for the same reason as the approach of extreme-value statistics: the model has to be used for conditions far beyond those used for model development and calibration. In this study the HBV model, a conceptual runoff model, was applied to four different catchments and differential split-sample testing (calibration on years with lower runoff peaks and testing it on years with higher peaks) was used to evaluate model performance for the situation when the model has to be used to simulate runoff during conditions different from those observed during calibration. To assess the value of improved calibration different goodness-of-fit measures were used, which allowed to explicitly consider the ability of the model to simulate groundwater-levels and peak flows. The results indicated that applying a model to conditions different from those during the calibration period might not give accurate results and that improved calibration procedures might not automatically provide more accurate flood estimations.
The International Commission for the Hydrology of the Rhine basin (CHR) has carried out a research project to assess the impact of climate change on the river flow conditions in the Rhine basin. Along a bottom-up line, different detailed hydrological models with hourly and daily time steps have been developed for representative sub-catchments of the Rhine basin. Along a top-down line, a water balance model for the entire Rhine basin has been developed, which calculates monthly discharges and which was tested on the scale of the major tributaries of the Rhine. Using this set of models, the effects of climate change on the discharge regime in different parts of the Rhine basin were calculated using the results of UKHI and XCCC GCM-experiments. All models indicate the same trends in the changes: higher winter discharge as a result of intensified snow-melt and increased winter precipitation, and lower summer discharge due to the reduced winter snow storage and an increase of evapotranspiration. When the results are considered in more detail, however, several differences show up. These can firstly be attributed to different physical characteristics of the studied areas, but different spatial and temporal scales used in the modelling and different representations of several hydrological processes (e.g., evapotranspiration, snow melt) are responsible for the differences found as well. Climate change can affect various socio-economic sectors. Higher temperatures may threaten winter tourism in the lower winter sport areas. The hydrological changes will increase flood risk during winter, whilst low flows during summer will adversely affect inland navigation, and reduce water availability for agriculture and industry. Balancing the required actions against economic cost and the existing uncertainties in the climate change scenarios, a policy of `no-regret and flexibility' in water management planning and design is recommended, where anticipatory adaptive measures in response to climate change impacts are undertaken in combination with ongoing activities.
Predictions of probabilities and magnitudes of extreme events are essential for water management. One approach for flood estimation is the use of conceptual runoff models. This approach, however, can be questioned for the same reason as the approach of extreme-value statistics: the model has to be used for conditions far beyond those used for model development and calibration. In this study the HBV model, a conceptual runoff model, was applied to four different catchments and differential split-sample testing (calibration on years with lower runoff peaks and testing it on years with higher peaks) was used to evaluate model performance for the situation when the model has to be used to simulate runoff during conditions different from those observed during calibration. To assess the value of improved calibration different goodness-of-fit measures were used, which allowed to explicitly consider the ability of the model to simulate groundwater-levels and peak flows. The results indicated that applying a model to conditions different from those during the calibration period might not give accurate results and that improved calibration procedures might not automatically provide more accurate flood estimations.
A parameterization is developed for the calculation of
evapotranspiration in three-dimensional atmospheric models. It
distinguishes separately between evaporation from the ground and
evapotranspiration from plant foliage. Soil water is stored in an active
layer of 1 m depth and a 10 cm surface layer is separately
distinguished. The evaporation from this soil is parameterized using a
high resolution multilayer model for comparison.
This parameterized evaporation from the soil is defined by either the
potential evaporation rate or by the maximum rate at which water can
diffuse to the surface, depending on which rate is smaller. The maximum
rate is obtained empirically in terms of various soil-hydraulic
parameters.
The evapotranspiration from plants occurs either as the evaporation of
water stored on the surface of the foliage or as the transpiration of
water extracted by roots from the soil. The flux of water from the outer
surface of foliage to the atmosphere above the canopy is determined by
the decrease in water vapor concentration from the foliage surface to
the overlying atmosphere and by the resistance of the foliage molecular
boundary layers and the bulk aerodynamic resistance of the canopy.
Transpired water encounters an additional stomatal resistance in passing
from the inside to the outside of leaves. The foliage temperature and
saturation vapor pressure are calculated from a model of the plant
canopy energy balance. Soil moisture determines the maximum rate at
which roots can extract water from the soil, and if the transpiration
demand exceeds this maximum rate, stomatal closure occurs until the
demand matches the root supply.
A parameterization of land evapotranspiration at the level of detail
described in this paper may be required to obtain a realistic diurnal
cycle of surface temperature and evapotranspiration for use in mesoscale
or global climate models. However, application of the developed
procedures is made difficult by the extreme complexity and small-scale
detail of surface processes, the lack of adequate data sets for surface
parameters, and the need for satisfactory parameterizations of other
components of. GCMs such as their rainfall and planetary boundary layer
treatments. Because of these difficulties, the development of validated
models of land surface processes first requires detailed sensitivity
studies to establish what further data sets are most urgently required
and what model improvement should be given highest priority.
This paper examines the possible solutions that may allow a rainfall-runoff model to cope with the existence of unknown intercatchment groundwater flows over a given catchment. On the basis of a large catchment set we compare four versions of the GR4J and the SMAR rainfall-runoff models that differ in the way they use one of their parameters to adjust catchment-scale water balance. We show that from both the hydrological likelihood and the modeling efficiency point of view it is preferable to explicitly represent intercatchment groundwater transfers. The surrogate corrective solutions tested in this paper (correcting or scaling factors applied to the climatic input data or to the catchment area) that are sometimes used in practice could be used on the sole grounds of streamflow simulation efficiency, but we show that they should be avoided since they may lead to obviously unrealistic corrections and consequently yield a similarly unrealistic distribution between evaporation streamflow and underground fluxes.
Extensions to the land surface scheme (LSS) in the National Centers for Environmental Prediction, regional, coupled, land-atmosphere weather prediction model, known as the mesoscale Eta model, are proposed and tested off-line in uncoupled mode to account for seasonal freezing and thawing of soils and snow-accumulation-ablation processes. An original model assumption that there is no significant heat transfer during redistribution of liquid water was relaxed by including a source/sink term in the heat transfer equation to account for latent heat during phase transitions of soil moisture. The parameterization uses the layer-integrated form of heat and water diffusion equations adopted by the original Eta-LSS. Therefore it simulates the total ice content of each selected soil layer. Infiltration reduction under frozen ground conditions was estimated by probabilistic averaging of spatially variable ice content of the soil profile. Off-line uncoupled tests of the new and original Eta-LSS were performed using experimental data from Rosemount, Minnesota. Simulated soil temperature and unfrozen water content matched observed data reasonably well. Neglecting frozen ground processes leads to significant underestimation/overestimation of soil temperature during soil freezing/thawing periods and underestimates total soil moisture content after extensive periods of soil freezing.
A new technique for determining the volumetric unfrozen water content of frozen soils is reported, which uses time domain reflectometry (TDR) to measure the dielectric property. Using precise temperature control, the technique, which was developed previously by others for unfrozen soils, has been successfully applied to the measurement of unfrozen water contents of frozen soils. Curves of the dielectric property versus temperature show a close similarity to unfrozen water content curves, for a variety of soils. Results from experiments on ice–water mixtures and from combined TDR–dilatometry experiments on frozen soils suggest that an empirical relationship obtained by Topp, Davis, and Annan may be applicable to frozen media as well as unfrozen soils. Using this relationship, dielectric values were converted to unfrozen water content values, and the results agreed very closely with published data for similar soils, determined by other methods. For silt loams, agreement is typically within ± 1½% in volumetric water content, and for clays ± 3 %. Some of this difference is undoubtedly due to soil sample variations.
Values of evapotranspiration are required for a variety of water planning activities in arid and semi-arid climates, yet data requirements are often large, and it is costly to obtain this information. This work presents a method where a few, readily available data (temperature, elevation) are required to estimate potential evapotranspiration (PET). A method using measured temperature and the calculated ratio of total to vertical radiation (after the work of Behnke and Maxey, 1969) to estimate monthly PET was applied for the months of April–October and compared with pan evaporation measurements. The test area used in this work was in Nevada, which has 124 weather stations that record sufficient amounts of temperature data. The calculated PET values were found to be well correlated (R2=0·940–0·983, slopes near 1·0) with mean monthly pan evaporation measurements at eight weather stations.In order to extrapolate these calculated PET values to areas without temperature measurements and to sites at differing elevations, the state was divided into five regions based on latitude, and linear regressions of PET versus elevation were calculated for each of these regions. These extrapolated PET values generally compare well with the pan evaporation measurements (R2=0·926–0·988, slopes near 1·0). The estimated values are generally somewhat lower than the pan measurements, in part because the effects of wind are not explicitly considered in the calculations, and near-freezing temperatures result in a calculated PET of zero at higher elevations in the spring months. The calculated PET values for April–October are 84–100% of the measured pan evaporation values. Using digital elevation models in a geographical information system, calculated values were adjusted for slope and aspect, and the data were used to construct a series of maps of monthly PET. The resultant maps show a realistic distribution of regional variations in PET throughout Nevada which inversely mimics topography. The general methods described here could be used to estimate regional PET in other arid western states (e.g. New Mexico, Arizona, Utah) and arid regions world-wide (e.g. parts of Africa). Copyright © 1999 John Wiley & Sons, Ltd.
A modified concept of hydrological response units (HRUs) for regional modelling of river basins using the PRMS/MMS model is presented. The HRUs are delineated by geographical information system (GIS) analysis from physiographic basin properties such as topography, soils, geology, rainfall and land use using a thorough hydrological systems analysis. The HRUs, once classified by GIS analysis, preserve the three-dimensional heterogeneity of the drainage basin. The River Bröl basin (A = 216 km2), Rheinisches Schiefergebirge, Germany was selected to apply the concept. In total, 23 HRUs were delineated and tested with the PRMS/MMS model using a 20-year hydrometeorological daily database. The hydrological systems analysis revealed that interflow is the dominant flow process through the basin's slopes and the major contribution to groundwater recharge and river runoff. This was accounted for by parameterizing the HRUs in the model control file to drain their surplus water not used for satisfying the demand of evapotranspiration to a common conceptual subsurface storage. This storage was simulated by interflow drainage to the groundwater aquifer in the valley floor, which in turn drained to the channel network. The PRMS/MMS model simulated the observed daily discharge very well and the fit was described by a daily correlation coefficient of r = 0.91. The NASIM and HSPF models using different means to represent the basin's physiographic heterogeneity were applied to the Bröl basin as well, but did not achieve this correlation. The HRU concept was found to be a reliable method for regional hydrological basin modelling and allows spatial up- and downscaling. Future research on this concept will focus on incorporating the variable precipitation distribution into the classification of HRUs and on the hydrodynamic routing of the modelled discharge. Additionally, satellite imagery must be used for classifying land use in macroscale drainage basins.
There are several indications that changes in land cover have influenced the hydrological regime of various river basins. In addition, the effects of climate change on the hydrological cycle and on the runoff behaviour of river catchments have been discussed extensively in recent years. However, it is at present rather uncertain how, how much and at which spatial scale these environmental changes are likely to affect the generation of storm runoff, and consequently the flood discharges of rivers. Firstly, this paper gives an overview of the possible effects of climatic and land-use change on storm runoff generation. Secondly, it discusses models dealing with the hydrological response to climate and land-use variations, including both the downscaling of climate information from global circulation models and the way flood forecasting models represent land-use conditions. Finally, two modelling studies of meso-scale catchments in Germany are presented: the first shows the possible impacts of climate change on storm runoff production, and the second the impacts of land-use changes. Copyright © 2002 John Wiley & Sons, Ltd.
Hydrological models must be reliable and robust as these qualities influence all applications based on model output. Previous studies on conceptual rainfall–runoff models have shown that one of the root causes of their output uncertainty is model over-parameterisation. The problem of poorly defined parameters has attracted much attention but has not yet been satisfactorily solved. We believe that the most fruitful way forward is to improve the structures where these parameters act. The main objective of this paper is to examine the role of complexity in hydrological models by studying the relation between the number of optimised parameters and model performance. An extensive comparative performance assessment of the structures of 19 daily lumped models was carried out on 429 catchments, mostly in France but also in the United States, Australia, the Ivory Coast and Brazil. Bulk treatment of the data showed that the complex models outperform the simple ones in calibration mode but not in verification mode. We argue that the main reason why complex models lack stability is that the structure, i.e. the way components are organised, is not suited to extracting information available in hydrological time-series. An inadequate complexity typically results in model over-parameterisation and parameter uncertainty. Although complexity has been used as a response to the challenge of predicting the hydrological effects of environmental changes, this study suggests that such models may have been developed with excessive confidence and that they could face difficulties of parameter estimation and structure validation when confronted with hydro-meteorological time-series. This comparative study indicates that some parsimonious models can yield promising results and should be further developed, although they are not able to tackle all types of problems, which would be the case if their complexity were ideally adapted.
A conceptual rainfall-runoff model was applied to 95 catchments in the Rhine basin for the purpose of modeling of the effect of land use change on the runoff. An approach to calibrate the model by associating the model parameters with the physical catchment characteristics was implemented. Land use, soil type, catchment size, and topographic structure were used as the bases for regionalization of the model parameters. Parameter values were initially associated with these catchment characteristics and were finally transferred into their catchment scale values using a transfer function, whose form was assumed a priori. A simultaneous model calibration was performed for a number of catchments with contrasting catchment characteristics. An optimization algorithm was used for the calibration with an objective function defined as the normalized sum of the squared differences between the modeled and the observed runoff. The regionalized model thus obtained was then used to model the resulting runoff for different land use scenarios generated in the model area. The results obtained suggest that increased urbanization leads to an increase in the lower peak runoff resulting from summer storm, while the increase in the higher peaks resulting from winter rainfall is very little. On the other hand, a considerable reduction of both the peak runoff and the total runoff volume resulted from intensified afforestation.
The prediction of extreme hydrological events in mesoscale catchments has been a main concern in hydrology because of their considerable societal impacts and because of the compelling evidence that anthropogenic activities significantly modify their occurrence likelihood. In this paper, nonlinear generalized models were used to predict extreme runoff characteristics like the specific volume, the frequency of high-flows, and the total drought duration. Explanatory variables included many physiographic, land cover, and climatic characteristics such as mean slope, aspect, elevation, type of geological formations, shares of a given land cover type, and many composed indicators relating antecedent precipitation index and atmospheric circulation patterns. All time-dependent variables were estimated semiannually for each subcatchment. The proposed method was tested in 46 subcatchments belonging to the Upper Neckar River basin covering an area of approximately 4000 km2 during the period from 1961 to 1993. The results of this study indicated that macro circulation patters derived from either subjective or operational classifications combined with other explanatory variables can be effectively used to predict seasonal extreme runoff characteristics at the mesoscale. Moreover, the results indicated that most runoff characteristics exhibited a distributional element other than normal and that the selection of nonlinear generalized models was an appropriate choice to deal with the heteroscedasticity of model errors.
In order to achieve a process-oriented simulation of hydrological processes in a meso-scale basin (101–103 km2), the spatially and temporally variable basin inputs (precipitation and energy) and runoff generation processes need to be adequately addressed by the model. The catchment model TACD (tracer aided catchment model, distributed) is based on experimental results including tracer studies at the mountainous Brugga basin (40 km2). This raster-based model (50×50 m2) works on an hourly basis, thus capturing the spatially and temporally variable inputs and processes. The model contains a process-realistic description of the runoff generation mechanism, which is based on a spatial delineation of eight units with the same dominating runoff generation processes. This defines the model structure and enables efficient model parameterisation. The model uses linear and non-linear reservoir routines to conceptualise runoff generation processes, and includes a routing routine (kinematic wave approach) to simulate surface runoff.The model is successfully applied to a 1-year period following minimal calibration (model efficiency 0.94). In addition, the runoff from both an independent 3-year period for the Brugga basin and a sub-basin (15.2 km2) is modelled well (model efficiencies 0.80 and 0.85, respectively) without re-calibration. The use of tracer data (i.e. dissolved silica) measured in outlet discharge demonstrates that the temporal mixing pattern of different runoff components is modelled correctly (multiple-response validation). The results show that a validated process-based model that correctly simulates the origin of runoff components and flow pathways must be the basis for integrating solute transport modelling of non-conservative species. Such a model can serve as tool to make predictions and test hypotheses about the first-order controls on hydrological responses.
The paper intends to review the current status of low-flow hydrology — a discipline which deals with minimum flow in a river during the dry periods of the year. The discussion starts with the analysis of low-flow generating mechanisms operating in natural conditions and the description of anthropogenic factors which directly or indirectly affect low flows. This is followed by the review of existing methods of low-flow estimation from streamflow time-series, which include flow duration curves, frequency analysis of extreme low-flow events and continuous low-flow intervals, baseflow separation and characterisation of streamflow recessions. The paper describes the variety of low-flow characteristics (indices) and their applications. A separate section illustrates the relationships between low-flow characteristics. The paper further focuses on the techniques for low-flow estimation in ungauged river catchments, which include a regional regression approach, graphical representation of low-flow characteristics, construction of regional curves for low-flow prediction and application of time-series simulation methods. The paper presents a summary of recent international low-flow related research initiatives. Specific applications of low-flow data in river ecology studies and environmental flow management as well as the problem of changing minimum river flows as the result of climate variability are also discussed. The review is largely based on the research results reported during the last twenty years.
This essay discusses some of the issues involved in the identification and predictions of hydrological models given some calibration data. The reasons for the incompleteness of traditional calibration methods are discussed. The argument is made that the potential for multiple acceptable models as representations of hydrological and other environmental systems (the equifinality thesis) should be given more serious consideration than hitherto. It proposes some techniques for an extended GLUE methodology to make it more rigorous and outlines some of the research issues still to be resolved.
The objective in this study is to explore a solution to the question whether model input data having higher spatial resolution and higher model resolution, as most people assume, lead to better model performance within a given modelling objective. An attempt was made to modify the conceptual rainfall–runoff model HBV to incorporate a spatially distributed structure. Additionally, three more model structures based on the HBV model concept were designed: lumped, semi-lumped and semi-distributed. An automatic calibration procedure based on simulated annealing optimization algorithm was followed for maximizing an objective function composed of Nash–Sutcliffe coefficients of several temporal aggregation steps. The predictive performance from each model was then assessed and compared with other model structures with respect to stream flow prediction at the catchment outlet. The spatial variation of the meteorological input was produced using external drift kriging method from available limited point measurements. The models were applied to a mesoscale catchment located in central Europe. The simulated hydrographs obtained using different model structures were analyzed through comparison of their Nash–Sutcliffe coefficients and other goodness-of-fit indices. For the present study, semi-distributed and semi-lumped model structures outperformed the distributed and fully-lumped model structures. A possible explanation why the distributed model did not perform better than the simpler model structures is the use of limited available spatial information. The models use interpolated precipitation and temperature as input, which probably cannot reflect the true spatial variability. Another possible explanation is that only discharge at the catchment outlet was predicted; which is the purpose for which lumped and semi-distributed models were actually designed.
This paper summarizes results from the Distributed Model Intercomparison Project (DMIP) study. DMIP simulations from twelve different models are compared with both observed streamflow and lumped model simulations. The lumped model simulations were produced using the same techniques used at National Weather Service River Forecast Centers (NWS-RFCs) for historical calibrations and serve as a useful benchmark for comparison. The differences between uncalibrated and calibrated model performance are also assessed. Overall statistics are used to compare simulated and observed flows during all time steps, flood event statistics are calculated for selected storm events, and improvement statistics are used to measure the gains from distributed models relative to the lumped models and calibrated models relative to uncalibrated models. Although calibration strategies for distributed models are not as well defined as strategies for lumped models, the DMIP results show that some calibration efforts applied to distributed models significantly improve simulation results. Although for the majority of basin-distributed model combinations, the lumped model showed better overall performance than distributed models, some distributed models showed comparable results to lumped models in many basins and clear improvements in one or more basins. Noteworthy improvements in predicting flood peaks were demonstrated in a basin distinguishable from other basins studied in its shape, orientation, and soil characteristics. Greater uncertainties inherent to modeling small basins in general and distinguishable inter-model performance on the smallest basin (65 km2) in the study point to the need for more studies with nested basins of various sizes. This will improve our understanding of the applicability and reliability of distributed models at various scales.
This paper presents a distributed model that is in operational use for forecasting flash floods in northern Austria. The main challenge in developing the model was parameter identification which was addressed by a modelling strategy that involved a model structure defined at the model element scale and multi-source model identification. The model represents runoff generation on a grid basis and lumped routing in the river reaches. Ensemble Kalman Filtering is used to update the model states (grid soil moisture) based on observed runoff. The forecast errors as a function of forecast lead time are evaluated for a number of major events in the 622 km2 Kamp catchment and range from 10% to 30% for 4–24 h lead times, respectively.
This study presents a grid-based modification of the HBV model concept and four regionalisation approaches using widely available catchment characteristics in the meso-scale Neckar catchment. The HBV model was adapted to allow for the simulation of catchment runoff and daily groundwater recharge in a high spatial discretisation. The resulting large number of model parameters requires the use of a regionalisation method which also ensures consistent parameter estimation. Therefore, in the first approach, functional relationships between catchment characteristics and model parameters have been defined a priori. These established relationships were used to calibrate the model by modifying the parameters of the transfer functions instead of the model parameters themselves. The results are compared to relationships derived from simultaneously calibrated model parameters constrained to form a function of catchment characteristics by a modification of the Lipschitz condition, a monotony condition and a combination of both constraints. Through this reduction of the available parameter space for optimisation, the problem of equifinality is avoided which often results in weak regression relationships between model parameters and catchment characteristics. The methodology is demonstrated using six subcatchments of the Neckar basin to set up the relationships and 51 other subcatchments to evaluate its performance. All four methods were able to produce reasonable parameter sets for most of the regionalisation catchments. As expected, all four methods failed to reproduce the observed discharge in karstic areas and in heavily modified or regulated river basins, which indicates their sensitivity to catchment characteristics. The modified Lipschitz condition produced the most efficient simulations of observed discharges in the regionalisation at the cost of some inconsistencies in the physical interpretation of the resulting relationships. The monotony condition preserved the assumed trends in the functions between cell properties and model parameters but produced sharp jumps which are not considered plausible. The combination of both methods seems to be the most promising because it produced equally good regionalisation results with much more consistent regression relationships. The approach can reproduce derived trends and the resulting relationships match our understanding of how the underlying processes are represented in the model.
This study is focused on analyses of scale dependency of lumped hydrological models with different formulations of the infiltration processes. Three lumped hydrological models of differing complexity were used in the study: the SAC-SMA model, the Oregon State University (OSU) model, and the simple water balance (SWB) model. High-resolution (4×4 km) rainfall estimates from the next generation weather radar (NEXRAD) Stage III in the Arkansas-Red river basin were used in the study. These gridded precipitation estimates are a multi-sensor product which combines the spatial resolution of the radar data with the ground truth estimates of the gage data. Results were generated from each model using different resolutions of spatial averaging of hourly rainfall. Although all selected models were scale dependent, the level of dependency varied significantly with different formulations of the rainfall-runoff partitioning mechanism. Infiltration-excess type models were the most sensitive. Saturation-excess type models were less scale dependent. Probabilistic averaging of the point processes reduces scale dependency, however, its effectiveness varies depending on the scale and the spatial structure of rainfall.
The principles governing the application of the conceptual model
technique to river flow forecasting are discussed. The necessity for a
systematic approach to the development and testing of the model is
explained and some preliminary ideas suggested.
In this paper, we analyze how our evaluation of the capacity of a rainfall-runoff model to represent low or high flows depends on the objective function used during the calibration process. We present a method to combine models to produce a more satisfactory streamflow simulation, on the basis of two different parameterizations of the same model. Where we previously had to choose between a more efficient simulation for either high flows or low flows (but inevitably less efficient in the other range), we show that a balanced simulation can be obtained by using a seasonal index to weigh the two simulations, providing good efficiency in both low and high flows.