Abstract:Abstract. The ambiguous representation of hydrological processes has led to the
formulation of the multiple hypotheses approach in hydrological modeling,
which requires new ways of model construction. However, most recent studies
focus only on the comparison of predefined model structures or building a
model step by step. This study tackles the problem the other way around: we
start with one complex model structure, which includes all processes deemed
to be important for the catchment. Next, we create 13 addit… Show more
“…All other parameters were subjectively set, as their conceptual nature does not allow to link them directly to physical processes. However, their ranges were in line with other studies that explored the Fulda catchment using models 23,35 and field experimental approaches like tritium 36 .…”
Section: Methodssupporting
confidence: 89%
“…For more details see Jehn et al . 23 .Figure 1Location of Hesse in Germany ( A ), Location of the Fulda catchment in Hesse ( B ) (gauging station Kämmerzell) and separation of the catchment by height ( C ) and vegetation/land cover ( D ).…”
Tightly constraint parameter ranges are seen as an important goal in constructing hydrological models, a difficult task in complex models. However, many studies show that complex models are often good at capturing the behaviour of a river. Therefore, this study explores the trade-offs between tightly constrained parameters and the ability to predict hydrological signatures, that capture the behaviour of a river. To accomplish this we built five models of differing complexity, ranging from a simple lumped model to a semi-lumped model with eight spatial subdivisions. All models are built within the same modelling framework, use the same data, and are calibrated with the same algorithm. We also consider two different methods for the potential evapotranspiration. We found that that there is a clear trade-off along the axis of complexity. While the more simple models can constrain their parameters quite well, they fail to get the hydrological signatures right. It is the other way around for the more complex models. The method of evapotranspiration only influences the parameters directly related to it. This study highlights that it is important to focus not only on parametric uncertainty. Tightly constrained parameters can be misguiding as they give credibility to oversimplified model structures.
“…All other parameters were subjectively set, as their conceptual nature does not allow to link them directly to physical processes. However, their ranges were in line with other studies that explored the Fulda catchment using models 23,35 and field experimental approaches like tritium 36 .…”
Section: Methodssupporting
confidence: 89%
“…For more details see Jehn et al . 23 .Figure 1Location of Hesse in Germany ( A ), Location of the Fulda catchment in Hesse ( B ) (gauging station Kämmerzell) and separation of the catchment by height ( C ) and vegetation/land cover ( D ).…”
Tightly constraint parameter ranges are seen as an important goal in constructing hydrological models, a difficult task in complex models. However, many studies show that complex models are often good at capturing the behaviour of a river. Therefore, this study explores the trade-offs between tightly constrained parameters and the ability to predict hydrological signatures, that capture the behaviour of a river. To accomplish this we built five models of differing complexity, ranging from a simple lumped model to a semi-lumped model with eight spatial subdivisions. All models are built within the same modelling framework, use the same data, and are calibrated with the same algorithm. We also consider two different methods for the potential evapotranspiration. We found that that there is a clear trade-off along the axis of complexity. While the more simple models can constrain their parameters quite well, they fail to get the hydrological signatures right. It is the other way around for the more complex models. The method of evapotranspiration only influences the parameters directly related to it. This study highlights that it is important to focus not only on parametric uncertainty. Tightly constrained parameters can be misguiding as they give credibility to oversimplified model structures.
“…The top-down approach starts with a simple model and adds components which are accepted as long as they improve model performance (Fenicia et al, 2016). The bottom-up approach starts with a complex model and removes processes to identify the most significant ones (Chang et al, 2017;Jehn et al, 2018). Others build predefined model structures based on expert knowledge (Gharari et al, 2014;Viglione et al, 2018;Wrede et al, 2015), varying complexity hypotheses (Fenicia et al, 2014;Van Esse et al, 2013) or different combinations of the conceptual choices of existing hydrological models (Coxon et al, 2014;Krueger et al, 2010).…”
Choosing (an) adequate model structure(s) for a given purpose, catchment, and data situation is a critical task in the modeling chain. However, despite model intercomparison studies, hypothesis testing approaches with modular modeling frameworks, and continuous efforts in model development and improvement, there are still no clear guidelines for identifying a preferred model structure. By introducing a framework for Automatic Model Structure Identification (AMSI), we support the process of identifying (a) suitable model structure(s) for a given task. The proposed AMSI framework employs a combination of the modular hydrological model RAVEN and the heuristic global optimization algorithm dynamically dimensioned search (DDS). It is the first demonstration of a mixed-integer optimization algorithm applied to simultaneously optimize model structure choices (integer decision variables) and parameter values (continuous decision variables) in hydrological modeling. The AMSI framework is thus able to sift through a vast number of model structure and parameter choices for identifying the most adequate model structure(s) for representing the rainfall-runoff behavior of a catchment. We demonstrate the feasibility of the approach by reidentifying given model structures that produced a specific hydrograph and show the limits of the current setup via a real-world application of AMSI on 12 MOPEX catchments. Results show that the AMSI framework is capable of inferring feasible model structure(s) reproducing the rainfall-runoff behavior of a given catchment. However, it is a complex optimization problem to identify model structure and parameters simultaneously. The variance in the identified structures is high due to near equivalent diagnostic measures for multiple model structures, reflecting substantial model equifinality. Future work with AMSI should consider the use of hydrologic signatures, case studies with multiple types of observation data, and the use of mixed-integer multiobjective optimization algorithms.
“…We developed a lumped conceptual rainfall-runoff-model (Figure 4) using the Catchment Modelling Framework (Kraft et al, 2011) Version 1.4.1 (Kraft et al, 2018) for Python 3. Catchment Modelling Framework is a python based programing library to build hydrological models from building blocks (Jehn et al, 2018). This framework has been applied to a variety of catchments and is considered to describe the underlying hydrological processes sufficiently well (Jehn et al, 2018;Maier et al, 2017;Windhorst et al, 2014).…”
Section: Model Setupmentioning
confidence: 99%
“…Catchment Modelling Framework is a python based programing library to build hydrological models from building blocks (Jehn et al, 2018). This framework has been applied to a variety of catchments and is considered to describe the underlying hydrological processes sufficiently well (Jehn et al, 2018;Maier et al, 2017;Windhorst et al, 2014). The model structure represents the conceptual understanding of the rainfall-runoff processes reported by Jacobs, Timbe, et al (2018).…”
Complex and costly discharge measurements are usually required to calibrate hydrological models. In contrast, water level measurements are straightforward, and practitioners can collect them using a crowdsourcing approach. Here we report how crowdsourced water levels were used to calibrate a lumped hydrological model. Using six different calibration schemes based on discharge or crowdsourced water levels, we assessed the value of crowdsourced data for hydrological modeling. As a benchmark, we used estimated discharge from automatically measured water levels and identified 2,500 parameter sets that resulted in the highest Nash-Sutcliffe-Efficiencies in a Monte Carlo-based uncertainty framework (Q-NSE). Spearman-Rank-Coefficients between crowdsourced water levels and modeled discharge (CS-SR) or observed discharge and modeled discharge (Q-SR) were used as an alternative way to calibrate the model. Additionally, we applied a filtering scheme (F), where we removed parameter sets, which resulted in a runoff that did not agree with the water balance derived from measured precipitation and publicly available remotely sensed evapotranspiration data. For the Q-NSE scheme, we achieved a mean NSE of 0.88, while NSEs of 0.43 and 0.36 were found for Q-SR and CS-SR, respectively. Within the filter schemes, NSEs approached the values achieved with the discharge calibrated model (Q-SR F 0.7, CS-SR F 0.69). Similar results were found for the validation period with slightly better efficiencies. With this study we demonstrate how crowdsourced water levels can be effectively used to calibrate a rainfall-runoff model, making this modeling approach a potential tool for ungauged catchments.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
