Early assessments of the hydrological impacts of global warming suggested both an intensification of the global water cycle and an expansion of dry areas. Yet these alarming conclusions were challenged by a number of latter studies emphasizing the lack of evidence in observations and historical simulations, as well as the large uncertainties in climate projections from the fifth phase of the Coupled Model Intercomparison Project (CMIP5). Here several aridity indices and a two‐tier attribution strategy are used to demonstrate that a summer midlatitude drying has recently emerged over the northern continents, which is mainly attributable to anthropogenic climate change. This emerging signal is shown to be the harbinger of a long‐term drying in the CMIP5 projections. Linear trends in the observed aridity indices can therefore be used as observational constraints and suggest that the projected midlatitude summer drying was underestimated by most CMIP5 models. Mitigating global warming therefore remains a priority to avoid dangerous impacts on global water and food security.

Solar radiation management by stratospheric aerosol injection (SRM‐SAI) has been proposed as a possible method to counteract anthropogenic global warming, with climate models suggesting it could reduce substantially global temperature and associated impacts. Its effectiveness as simulated by Earth system models exhibits, however, large uncertainties, implying high risks for natural and human ecosystems. Here we identify an emergent relationship linking the long‐term global land surface cooling due to SRM‐SAI and the short‐term cooling following the twentieth century major volcanic eruptions across an Earth system models ensemble. This emergent relationship, combined with observations and reanalysis data, is used to constrain the global land surface temperature (LT) response to reduced downward solar radiation. Based on these constraints, we find a mean decrease in land surface temperature of 0.44 K·W−1·m2, 20% smaller than the unconstrained multimodel mean. This new estimate may affect how trade‐offs between cost, risk, and effectiveness of SRM‐SAI might be considered.

Solar radiation modification (SRM) is known to strengthen both land and ocean carbon uptake because of its impacts on surface temperature, solar radiation, and other potential drivers of the global carbon cycle. However, the magnitude and timing of the response of both land and ocean carbon uptake to SRM and its consequence on allowable CO2 emissions remain poorly understood. Here we use the results of six Earth system models simulating a continuous stratospheric injection of 5 Tg of sulfur dioxide per year between 2020 and 2069 under the representative concentration pathways 4.5 to investigate the impact of SRM on land and ocean carbon uptake. We find that 50 years of SRM under this protocol increases the allowable CO2 emissions by 40 ± 19 GtC; 85% of this additional uptake of carbon is stored in the land biosphere and 15% in the ocean. This increase in allowable CO2 emissions is however not sustainable after the stoppage of SRM. Earth system models predict a mean release of 8 ± 11 GtC of the carbon back to the atmosphere 20 years after the stoppage which is dominated by large uncertainties in the response of the simulated land carbon cycle to rising temperature and solar radiation. We demonstrate that the time scales of carbon dioxide removal (CDR) potential of SRM are smaller than the time scales of the geological storage assumed in well‐established CDR options. This shows that the CDR potential of SRM should be compared to well‐established CDR options with caution.