Impact of subjective modeling decisions on hydrological modeling

<p>Modelers often make different decisions in building hydrologic models based on their experience and modeling philosophy. Consequently, a wide range of models is developed, which differ in many aspects of conceptualization and implementation. This diversity of models has been useful to explore a myriad of scientific and applied questions, but it has also led to great confusion on choosing the appropriate model configurations in compliance with the dominant processes in the study area. Also, modeling decisions during model configuration introduce subjectivity from the modeler. To provide guidance to select the best-suited model configuration for a catchment it is required to examine and evaluate the different model representations of hydrological processes and their impact on model simulations. In this study, we show that modeling decisions during the model configuration, beyond the model choice, also impact the model results. The framework, Structure for Unifying Multiple Modeling Alternatives (SUMMA; Clark et al., 2015a, b) is used in this study to disentangle the model components which helps to have a controlled and systematic evaluation of multiple models representations. The area chosen for the study is the Malaprabha catchment in the Karnataka state of India. The impact of the choice of parameterizations and parameter values on the model simulations are shown. To improve upon the traditional model evaluation methods, hydrological signatures are made use to have a hydrologically meaningful evaluation of model simulations. This study helped to identify the suitable model configuration for the Malaprabha catchment. Multiple working hypotheses during model configuration which is possible with the help of such flexible framework like SUMMA can provide insights on the impact of subjective modeling decisions.</p>

Application of Systematical Optimization GM(1,1) Model Based on Lifting Wavelet in Dam Displacement Forecast

A new GM(1,1) model is proposed in the paper thinking about of the influence of the traditional modeling accuracy bias of subjective modeling and noise interfere. Using the new model, the noisy part of monitoring data could be deleted and prediction error could be decreased. At the same time, the constraint conditions of minimum fitting error sum of squares of each variable during once accumulation is raised to build and optimize a prediction model with optimal initial values based on the Principle of Least Squares. Moreover, the new GM(1,1) model is established by improving the background value and gray value with consideration of systematical optimization. Finally through the practical case, the results show that the calculation result of using the new GM(1,1) model is smaller and the accuracy is higher than using the traditional model.