Skew-normal distribution model for rainfall uncertainty estimation in a distributed hydrological model

Despite the progress made by numerous contributions in recent decades on uncertainty in hydrological simulation, there are still knowledge gaps in estimating uncertainty sources, especially associated with precipitation. The aim of this study was to determine the precipitation uncertainty through an error model based on the skew normal distribution function and to evaluate the effect of its propagation towards the simulated flow with the TETIS distributed hydrological model in a poorly instrumented tropical Andean basin. The results show the performance of the hydrological model is more sensitive to the location of the meteorological station used than to the number of stations employed in a real case with scarce information. Implementing the Bayesian approach for the study of uncertainty in input data such as precipitation is essential for its quantification, improving the knowledge of how this source of error propagates to the results of the hydrological simulation.