Although anthropogenic climate change is expected to have caused large shifts in temperature and rainfall, the detection of human influence on global drought has been complicated by large internal variability and the brevity of observational records. Here we address these challenges using reconstructions of the Palmer drought severity index obtained with data from tree rings that span the past millennium. We show that three distinct periods are identifiable in climate models, observations and reconstructions during the twentieth century. In recent decades (1981 to present), the signal of greenhouse gas forcing is present but not yet detectable at high confidence. Observations and reconstructions differ significantly from an expected pattern of greenhouse gas forcing around mid-century (1950-1975), coinciding with a global increase in aerosol forcing. In the first half of the century (1900-1949), however, a signal of greenhouse-gas-forced change is robustly detectable. Multiple observational datasets and reconstructions using data from tree rings confirm that human activities were probably affecting the worldwide risk of droughts as early as the beginning of the twentieth century.

From 2012 to 2016, California experienced one of the worst droughts since the start of observational records. As in previous dry periods, precipitation-inducing winter storms were steered away from California by a persistent atmospheric ridging system in the North Pacific. Here we identify a new link between Arctic sea-ice loss and the North Pacific geopotential ridge development. In a two-step teleconnection, sea-ice changes lead to reorganization of tropical convection that in turn triggers an anticyclonic response over the North Pacific, resulting in significant drying over California. These findings suggest that the ability of climate models to accurately estimate future precipitation changes over California is also linked to the fidelity with which future sea-ice changes are simulated. We conclude that sea-ice loss of the magnitude expected in the next decades could substantially impact California's precipitation, thus highlighting another mechanism by which human-caused climate change could exacerbate future California droughts.

Possible future climate changes in California are investigated from a varied set of climate change model simulations. These simulations, conducted by three state-of-the-art global climate models, provide trajectories from three greenhouse gas (GHG) emission scenarios. These scenarios and the resulting climate simulations are not “predictions,” but rather are a limited sample from among the many plausible pathways that may affect California’s climate. Future GHG concentrations are uncertain because they depend on future social, political, and technological pathways, and thus the IPCC has produced four “families” of emission scenarios. To explore some of these uncertainties, emissions scenarios A2 (a medium-high emissions) and B1 (low emissions) were selected from the current IPCC Fourth climate assessment, which provides several recent model simulations driven by A2 and B1 emissions. The global climate model simulations addressed here were from PCM1, the Parallel Climate Model from the National Center for Atmospheric Research (NCAR) and U.S. Department of Energy (DOE) group, and CM2.1 from the National Oceanic and Atmospheric Administration (NOAA) Geophysical Fluids Dynamics Laboratory (GFDL). As part of the scenarios assessment, a statistical technique using properties of historical weather data was employed to correct model biases and “downscale” the global-model simulation of future climates to a finer level of detail, onto a grid of approximately 7 miles (12 kilometers), which is more suitable for impact studies at the scales needed by California decision makers. In current climate-change simulations, temperatures over California warm significantly during the twenty-first century, with temperature increases from approximately +3ºF (1.5ºC) in the lower emissions scenario (B1) within the less responsive model (PCM1) to +8ºF (4.5ºC) in the higher emissions scenario (A2) within the more responsive model (CM2.1). Three of the simulations (all except the low-emission scenario run of the low-response model) exhibit more warming in summer than in winter. In all of the simulations, most precipitation continues to occur in winter, with virtually all derived from North Pacific winter storms. Relatively little change in overall precipitation is projected. Climate warming has a profound influence in diminishing snow accumulations, because there is more rain and less snow, and earlier snowmelt. These snow losses increase as the warming increases, so that they are most severe under climate changes projected by the more sensitive model with the higher GHG emissions.

In this work, we compare the rate of warming of summertime extreme temperatures (summer maximum value of daily maximum temperature; TXx) relative to the local mean (summer mean daily maximum temperature; TXm) over the Northern Hemisphere in observations and one set of large ensemble (LE) simulations. During the 1979–2021 historical period, observations and simulations show robust warming trends in both TXm and TXx almost everywhere in the Northern Hemisphere, except over the eastern U.S. where observations show a slight cooling trend in TXx, which may be a manifestation of internal variability. We find that the observed warming rate in TXx is significantly smaller than in TXm in North Africa, western North America, Siberia, and Eastern Asia, whereas the warming rate in TXx is significantly larger over the Eastern U.S., the U.K., and Northwestern Europe. This observed geographical pattern is successfully reproduced by the vast majority of the LE members over the historical period, and is persistent (although less intense) in future climate projections over the 2051–2100 period. We also find that these relative warming patterns are mostly driven by the local hydroclimate conditions. TXx warms slower than TXm in the hyper-arid, arid, semi-arid and moist regions, where trends in the partitioning of the turbulent surface fluxes between the latent and sensible heat flux are similar during regular and extreme hot days. In contrast, TXx warms faster than TXm in dry-subhumid regions where trends in the partitioning of the surface fluxes are significantly different between regular and extreme hot days, with a larger role of sensible heat flux during the extreme hot days.