Progress in hydrological science strongly relies on meaningful hypothesis testing. Conceptual hydrological models are often developed based on qualitative perceptions of how a catchment works, due to the lack of direct observations on internal catchment processes. As a result, model realism is often judged a posteriori, based on how well model predictions agree with available data. Model calibration requires making assumptions about the nature of the errors, which are seldom based on a priori information, and require posterior scrutiny. A meaningful evaluation of the realism of different assumptions influences the ability to distinguish between different hypotheses, and critically affects the inference of catchment processes.
The purpose of this session is to make the point on “where are we at” in hypothesis testing and what can we do to improve it. Recent research developments have shown the importance of different diagnostic approaches, including statistical and process oriented techniques, which substantiate a critical scrutiny of different model hypotheses. On the modelling side, new research developments have shown the importance of dedicating attention to different aspects of model design, including numerical implementation and solution schemes. Multi-model approaches expanded the spectrum of available hydrological models and enabled more meaningful comparison. This session welcomes presentations of different topics with a common objective of illustrating and enhancing hypothesis testing in hydrology.