Human activity has a profound influence on river discharges, hydrological extremes and water-related hazards. In this study, we compare the results of five state-of-the-art global hydrological models (GHMs) with observations to examine the role of human impact parameterizations (HIP) in the simulation of mean, high-and low-flows. The analysis is performed for 471 gauging stations across the globe for the period 1971-2010. We find that the inclusion of HIP improves the performance of the GHMs, both in managed and near-natural catchments. For near-natural catchments, the improvement in performance results from improvements in incoming discharges from upstream managed catchments. This finding is robust across the GHMs, although the level of improvement and the reasons for it vary greatly. The inclusion of HIP leads to a significant decrease in the bias of the long-term mean monthly discharge in 36%-73% of the studied catchments, and an improvement in the modeled hydrological variability in 31%-74% of the studied catchments. Including HIP in the GHMs also leads to an improvement in the simulation of hydrological extremes, compared to when HIP is excluded. Whilst the inclusion of HIP leads to decreases in the simulated high-flows, it can lead to either increases or decreases in the low-flows. This is due to the relative importance of the timing of return flows and reservoir operations as well as their associated uncertainties. Even with the inclusion of HIP, we find that the model performance is still not optimal. This highlights the need for further research linking human management and hydrological domains, especially in those areas in which human impacts are dominant. The large variation in performance between GHMs, regions and performance indicators, calls for a careful selection of GHMs, model components and evaluation metrics in future model applications.

Global-scale hydrological models are routinely used to assess water scarcity, flood hazards and droughts worldwide. Recent efforts to incorporate anthropogenic activities in these models have enabled more realistic comparisons with observations. Here we evaluate simulations from an ensemble of six models participating in the second phase of the Inter-Sectoral Impact Model Inter-comparison Project (ISIMIP2a). We simulate monthly runoff in 40 catchments, spatially distributed across eight global hydrobelts. The performance of each model and the ensemble mean is examined with respect to their ability to replicate observed mean and extreme runoff under human-influenced conditions. Application of a novel integrated evaluation metric to quantify the models' ability to simulate timeseries of monthly runoff suggests that the models generally perform better in the wetter equatorial and northern hydrobelts than in drier southern hydrobelts. When model outputs are temporally aggregated to assess mean annual and extreme runoff, the models perform better. Nevertheless, we find a general trend in the majority of models towards the overestimation of mean annual runoff and all indicators of upper and lower extreme runoff. The models struggle to capture the timing of the seasonal cycle, particularly in northern hydrobelts, while in southern hydrobelts the models struggle to reproduce the magnitude of the seasonal cycle. It is noteworthy that over all hydrological indicators, the ensemble mean fails to perform better than any individual model-a finding that challenges the commonly held perception that model ensemble © 2018 The Author(s). Published by IOP Publishing Ltd Environ. Res. Lett. 13 (2018) 065015 estimates deliver superior performance over individual models. The study highlights the need for continued model development and improvement. It also suggests that caution should be taken when summarising the simulations from a model ensemble based upon its mean output.

We present one of the first climate change impact assessments on river runoff that utilises an ensemble of global hydrological models (Glob-HMs) and an ensemble of catchment-scale hydrological models (Cat-HMs), across multiple catchments: the upper Amazon, Darling, Ganges, Lena, upper Mississippi, upper Niger, Rhine and Tagus. Relative changes in simulated mean annual runoff (MAR) and four indicators of high and low extreme Climatic Change (2017) 141:577-595 DOI 10.1007/s10584-016-1773 The original version of this article was revised: The middle initial "J." which had erroneously been added to the name of second author Jamal Zaherpour has now been taken out. Centre for International Development and Environmental Research (ZEU), Justus Liebig University, Giessen, Germany flows are compared between the two ensembles. The ensemble median values of changes in runoff with three different scenarios of global-mean warming (1, 2 and 3°C above preindustrial levels) are generally similar between the two ensembles, although the ensemble spread is often larger for the Glob-HM ensemble. In addition the ensemble spread is normally larger than the difference between the two ensemble medians. Whilst we find compelling evidence for projected runoff changes for the Rhine (decrease), Tagus (decrease) and Lena (increase) with global warming, the sign and magnitude of change for the other catchments is unclear. Our model results highlight that for these three catchments in particular, global climate change mitigation, which limits global-mean temperature rise to below 2°C above preindustrial levels, could avoid some of the hydrological hazards that could be seen with higher magnitudes of global warming.

Importance of evaluation of global hydrological models (gHMs) before doing climate impact assessment was underlined in several studies. The main objective of this study is to evaluate the performance of six gHMs in simulating observed discharge for a set of 57 large catchments applying common metrics with thresholds for the monthly and seasonal dynamics and summarize them estimating an aggregated index of model performance for each model in each basin. One model showed a good performance, and other five showed a weak or poor performance in most of the basins. In 15 catchments, evaluation results of all models were poor. The model evaluation was supplemented by climate impact assessment for a subset of 12 representative catchments using (1) usual ensemble mean approach and (2) weighted mean approach based on model performance, and the outcomes were compared. The comparison of impacts in terms of mean monthly and mean annual discharges using two approaches has shown that in four basins, differences were negligible or small, and in eight catchments, differences in mean monthly, mean annual discharge or both were moderate to large. The spreads were notably decreased in most cases when the second method was applied. It can be concluded that for improving credibility of projections, the model evaluation and application of the weighted mean approach could be recommended, especially if the mean monthly (seasonal) impacts are of interest, whereas the ensemble mean approach could be applied for projecting the mean annual changes. The calibration of gHMs could improve their performance and, consequently, the credibility of projections.

This study presents a novel application of machine learning to deliver optimised, multi-model combinations (MMCs) of Global Hydrological Model (GHM) simulations. We exemplify the approach using runoff simulations from five GHMs across 40 large global catchments. The benchmarked, median performance gain of the MMC solutions is 45% compared to the best performing GHM and exceeds 100% when compared to the EM. The performance gain offered by MMC suggests that future multimodel applications consider reporting MMCs, alongside the EM and intermodal range, to provide endusers of GHM ensembles with a better contextualised estimate of runoff. Importantly, the study highlights the difficulty of interpreting complex, non-linear MMC solutions in physical terms. This indicates that a pragmatic approach to future MMC studies based on machine learning methods is required, in which the allowable solution complexity is carefully constrained.