Detection, attribution, and specifying mechanisms of hydrological changes in geographically different river basins

Our study is aimed at detection of directional trends in streamflow data observed in large rivers located in different climatic zones and attribution of the detected changes to climate drivers. We consider detection and attribution as interrelated study stages within a suggested hypothesis testing framework with the use of a hydrological model. First, we test the significance of the trends in the observed streamflow data series of 74 to 82 years long and evaluate the model's ability to reproduce the trends, so that the trends in the simulated data are statistically indistinguishable from the corresponding observed trends. Herewith, the model is forced by the reanalysis climate data. Then, for the basins where the model reproduces the trends, we move to the attribution stage of the study. At this stage, the hydrological model is forced by the counterfactual (detrended) climate data. If the trend is not detected in the counterfactual-climate-forced simulations, we conclude that the detected observed changes are likely to be attributed to the climate trend. The suggested testing procedure is applied for four river basins: Lena, Selenga, Vyatka, and Pechora. The corresponding hydrological models are developed on the basis of the ECOMAG modeling platform. We conclude that the detected trends in the observed annual flow data series for the Lena, Selenga, and Vyatka rivers, as well as the trends in high flow for the Lena and Selenga rivers, can be attributed to climate drivers with a high confidence. Regional differences in basin mechanisms governing the detected changes are analyzed.