Does a comprehensive evaluation increase confidence in the hydrological model intended for climate impact assessment?

The objective of the study is to verify a hypothesis that the hydrological model that successfully passed a comprehensive evaluation test is more suitable for impact study than the other model that failed the test. The hypothesis verification is carried out on an example of the physically-based hydrological models ECOMAG and SWAP, which are set up for the two great Arctic basins: the Lena and the Mackenzie rivers. Three versions of every model are compared: (1) the model with a priori assessed parameters (without any calibration); (2) the model calibrated against the streamflow observations at the basin outlets only, and (3) the model calibrated against the streamflow observations at several sites within the basins. The comprehensive evaluation procedure, which includes enhanced tests of model performance and robustness, is applied for all the versions of every model. The performance of the models is compared at multiple sites within the catchments and for multiple hydrological indicators of interest (high flow, low flow, multi-year trends). The robustness of the models is compared through statistical significance of the differences in the performance criteria of the model for climatically contrasting periods composed from the historical meteorological data. From the evaluation results, we identified the preferable (in terms of the assigned criteria) models and established the limits of the models applicability. Then all the compared models, being forced by the Global Climate Model ensemble data, were applied to simulate flow projections for the 21^st century and assess the projection uncertainty. The experiment demonstrates that the basin outlet flow projections simulated by the non-calibrated models differ from the projections of the calibrated models in terms of the mean ensemble trajectories and their uncertainty. Thus, under the study conditions (used models, studied basins), we answer "yes" to the question posed in the title of the presentation.