This study investigates the suitability of the asynchronous ensemble Kalman filter (AEnKF) and a partitioned updating scheme for hydrological forecasting. The AEnKF requires forward integration of the model for the analysis and enables assimilation of current and past observations simultaneously at a single analysis step. The results of discharge assimilation into a grid-based hydrological model (using a soil moisture error model) for the Upper Ourthe catchment in the Belgian Ardennes show that including past predictions and observations in the data assimilation method improves the model forecasts. Additionally, we show that elimination of the strongly non-linear relation between the soil moisture storage and assimilated discharge observations from the model update becomes beneficial for improved operational forecasting, which is evaluated using several validation measures.

This paper presents a hybrid local-global sensitivity analysis method termed the Distributed Evaluation of Local Sensitivity Analysis (DELSA), which is used here to identify important and unimportant parameters and evaluate how model parameter importance changes as parameter values change. DELSA uses derivative-based "local" methods to obtain the distribution of parameter sensitivity across the parameter space, which promotes consideration of sensitivity analysis results in the context of simulated dynamics. This work presents DELSA, discusses how it relates to existing methods, and uses two hydrologic test cases to compare its performance with the popular global, variance-based Sobol' method. The first test case is a simple nonlinear reservoir model with two parameters. The second test case involves five alternative "bucket-style" hydrologic models with up to 14 parameters applied to a medium-sized catchment (200 km²) in the Belgian Ardennes. Results show that in both examples, Sobol' and DELSA identify similar important and unimportant parameters, with DELSA enabling more detailed insight at much lower computational cost. For example, in the real-world problem the time delay in runoff is the most important parameter in all models, but DELSA shows that for about 20% of parameter sets it is not important at all and alternative mechanisms and parameters dominate. Moreover, the time delay was identified as important in regions producing poor model fits, whereas other parameters were identified as more important in regions of the parameter space producing better model fits. The ability to understand how parameter importance varies through parameter space is critical to inform decisions about, for example, additional data collection and model development. The ability to perform such analyses with modest computational requirements provides exciting opportunities to evaluate complicated models as well as many alternative models. Key Points Sobol' and DELSA methods showed that they yield similar sensitivity results DELSA provides more information at lower calculation costs than Sobol' DELSA is recommended as effective and efficient method for parameter sensitivity

This paper presents a study on the optimal setup for discharge assimilation within a spatially distributed hydrological model. The Ensemble Kalman filter (EnKF) is employed to update the grid-based distributed states of such an hourly spatially distributed version of the HBV-96 model. By using a physically based model for the routing, the time delay and attenuation are modelled more realistically. The discharge and states at a given time step are assumed to be dependent on the previous time step only (Markov property).

Synthetic and real world experiments are carried out for the Upper Ourthe (1600 km2), a relatively quickly responding catchment in the Belgian Ardennes. We assess the impact on the forecasted discharge of (1) various sets of the spatially distributed discharge gauges and (2) the filtering frequency. The results show that the hydrological forecast at the catchment outlet is improved by assimilating interior gauges. This augmentation of the observation vector improves the forecast more than increasing the updating frequency. In terms of the model states, the EnKF procedure is found to mainly change the pdfs of the two routing model storages, even when the uncertainty in the discharge simulations is smaller than the defined observation uncertainty.

Sound spatially distributed rainfall fields including a proper spatial and temporal error structure are of key interest for hydrologists to force hydrological models and to identify uncertainties in the simulated and forecasted catchment response. The current paper presents a temporally coherent error identification method based on time-dependent multivariate spatial conditional simulations, which are conditioned on preceding simulations. A sensitivity analysis and real-world experiment are carried out within the hilly region of the Belgian Ardennes. Precipitation fields are simulated for pixels of 10 km × 10 km resolution. Uncertainty analyses in the simulated fields focus on (1) the number of previous simulation hours on which the new simulation is conditioned, (2) the advection speed of the rainfall event, (3) the size of the catchment considered, and (4) the rain gauge density within the catchment. The results for a sensitivity analysis show for typical advection speeds >20 km hĝ̂'1, no uncertainty is added in terms of across ensemble spread when conditioned on more than one or two previous hourly simulations. However, for the real-world experiment, additional uncertainty can still be added when conditioning on a larger number of previous simulations. This is because for actual precipitation fields, the dynamics exhibit a larger spatial and temporal variability. Moreover, by thinning the observation network with 50%, the added uncertainty increases only slightly and the cross-validation shows that the simulations at the unobserved locations are unbiased. Finally, the first-order autocorrelation coefficients show clear temporal coherence in the time series of the areal precipitation using the time-dependent multivariate conditional simulations, which was not the case using the time-independent univariate conditional simulations. The presented work can be easily implemented within a hydrological calibration and data assimilation framework and can be used as an improvement over currently used simplistic approaches to perturb the interpolated point or spatially distributed precipitation estimates.