Using big data to detect and attribute global hydrological change

Anthropogenic climate change as well as human water and land management are expected to impact the world's freshwater resources and hydrological extremes. However, model estimates are imperfect, leaving many open questions regarding the sign and the magnitude of the associated impacts. Likewise, relevant in-situ observations have incomplete observational coverage and are often not sufficient to disentangle the effects of anthropogenic climate change and land management from natural variability and observational errors. The recently established Global Streamflow Indices and Metadata archive (GSIM), includes observations from more than 30000 stations and hence provides and unprecedented opportunity to investigate observed trends in global streamflow extremes and water balance indicators. Analysing the GSIM reveals complex spatial patterns of global hydrological change, where some regions (e.g. the Mediterranean) are getting significantly dryer, while others (e.g. North Asia) have been subject to wetting. Notably, these changes are consistent throughout the entire flow distribution, implying that increasing drought risk might be compensated by decreasing flood magnitudes. To test the hypothesis that the observed trends are caused by anthropogenic climate change or human land management detection and attribution methods are used. These methods exploit massive ensembles of factorial global model experiments in which the hypothesised agents of change (e.g. greenhouse gas emissions, human water management) are systematically switched on and off. Focussing on Europe, it is shown that the observed wetting (North) and drying (South) pattern can only be captured by state of the art climate model simulations if human greenhouse gas emissions are considered. Subsequently, world-wide observations will be subject to a detection and attribution analysis on the basis of a large ensemble of global models to disentangle the relative effects of anthropogenic climate change and human water management on trends in streamflow extremes and water balance indicators. The results highlight that that anthropogenic climate change is already detectable freshwater resources at continental to global scale, thereby increasing confidence in future projections of the world's freshwater resources.