A Robust calibration/validation protocol of a hydrological model using hidden Markov states

Abstract. The impacts of climate and land-use changes make the stationary assumption in hydrology obsolete. Moreover, there is still considerable uncertainty regarding the future evolution of the Earth's climate and the extent of the alteration of flow regimes. In that context, it is crucial to assess the performance of a hydrologic model over a wide range of climates and their corresponding hydrologic conditions. We propose a calibration/validation protocol based on the differential split-sample test and numerous, contrasted, climate sequences identified through a Hidden Markov Model (HMM) classification. The proposed protocol is tested on the Senegal River in West Africa. Results show that when the time series of river discharges does not exhibit a clear climate trend, or when it has multiple change points, classical rupture tests are useless and HMM classification is a viable alternative as long as the climate sub-sequences are long enough.