Novel methods for uncertainty analysis of hydrologic models

The uncertainty in model output means that forecasts should be considered in aprobabilistic way or using fuzzy logic. There are many examples in which deterministicapproaches are not sufficient due to the stochastic nature of input variables, for examplea deterministic forecast may fail when an extreme event occurs.The aim of this study was to explore the resampling methods for uncertainty analysis ofhydrologic models, including the Markov Chain Monte Carlo (MCMC) methodology,Metropolis algorithm in detail to estimate the uncertainties of simulated results. Thelimitations of this approach are investigated. Although this method does provideaccurate results, the large number of simulations is required, and with a complex modelrequiring a lot of time even for one simulation, it takes a long time. This clearlyindicates the requirement of time saving techniques, therefore, this study is also aimedto explore the methods of uncertainty analysis, so two algorithms, adaptive clustercovering (ACCO) and shuffled complex evolution metropolis (SCEM-UA) wereinvestigated.The investigation of MCMC method is grouped around the case studies: a knowndistribution case study, linear model case study and lumped conceptual rainfall-runofftank model case study for the Bagmati catchment in Nepal. Firstly, a two-dimensionalbanana-shaped posterior distribution is simulated to test the code and investigate theapplicability of the Metropolis-Hastings algorithm for assessment of parameteruncertainty for this two-dimensional highly nonlinear distribution. Secondly, linearmodel experiments are done, data sets with noise and without noise are applied toinvestigating the performance of Metropolis-Hastings (MH) algorithm, in order tounderstand the algorithm fully. ACCO and SCEM-UA are also used in these two cases.Thirdly, for Bagmati catchment case study, the mean daily data of hydrological stationon the Bagmati river at Pandheradobhan is used and totally 2000 days data from Jan1988 to June 1993 were collected to estimate the uncertainties of simulated resultsoriginating from both model parameters and all sources of uncertainty. ACCO was alsoinvestigated whether it can be used in uncertainty analysis and improve the simulationefficiency.The results obtained demonstrate that ACCO can give reasonable parameterdistributions for those parameters with steep shape and one peak, but it fails in othercases. MH can give the accurate results in most cases. It can be concluded that ACCOand SCEM-UA are more efficient than MH for calibration since they are able to obtainthe best solution at much faster speed than MH, whereas the MH explores the entireparameter space and can produce more reliable results than the other two methods