Wind and solar energy technologies are, by their nature, variable. Variations in resource availability, based on weather patterns, occur on intra-day to inter-annual time scales. Many energy system models optimize over a single year of input weather and electricity demand data. Energy system planners need increased understanding of the variability in generation potential across multiple years and how this could impact model results. A system achieving 100% reliability modeled using Year A data will not necessarily achieve 100% reliability when applied to Year B data unless an overbuild safety margin is added. We demonstrate: 1) model results can vary significantly based on the year of data used, 2) adding wind and solar does not necessarily reduce the predictability of meeting reliability targets year-to-year and can improve predictability in many cases, and 3) we illustrate a method to derive safety margins to predictably meet 100% reliability year after year and find the least-cost option.

Solar geoengineering has been suggested as a potential approach to counteract the anthropogenic global warming. Major volcanic eruptions have been used as natural analogues to large-scale deployments of stratospheric aerosol geoengineering, yet difference in climate responses to these forcings remains unclear. Among many factors characterizing the difference between the two, durations of the additional aerosol layer in the stratosphere differ substantially between volcanic eruptions and the SAI geoengineering. Sulfate aerosols from volcanic eruptions typically stay in the stratosphere for one to two years. Stratospheric aerosol geoengineering, however, if used to counteract anthropogenic warming, would need to be deployed quasi-continuously and thus the additional aerosols would stay in the stratosphere persistently. Using the NCAR CESM model, we compare the climate response to two highly idealized stratospheric aerosol forcings that have different durations: a short-term pulse representative of volcanic eruptions and a long-term sustained forcing representative of geoengineering. For the same amount of global mean cooling, the pulse case causes much larger reductions in surface temperature over land relative to the sustained case. This greater cooling over land leads to a larger decrease in the vertical motion of air over land in lower atmosphere, and reduces water vapor transport from the ocean to land. For similar amounts of global cooling, the decrease in land runoff caused by a short-term pulse aerosol forcing is about twice as large as that caused by a sustained aerosol forcing. Our results clearly demonstrate difference in the climate response to volcanic-like and geoengineering-like stratospheric aerosol forcings, and suggest that caution should be exercised when extrapolating results from volcanic eruptions to the SAI geoengineering deployments. However, observations and simulations of climate impacts from volcanic eruptions test many of the same physical mechanisms that would come into play in a stratospheric aerosol geoengineering scenario, and thus major volcanic eruptions remain as valuable analogues for solar geoengineering deployment.

Global mean surface air temperature (T) variability on subdecadal timescales can be of substantial magnitude relative to the long-term global warming signal and such variability has been associated with considerable environmental and societal impacts. Therefore, probabilistic foreknowledge of short-term T evolution may be of value for anticipating and mitigating some course-resolution climate-related risks. Here we present an empirically-based methodology that utilizes global spatial patterns of annual surface air temperature to predict subsequent annual T anomalies via Partial Least Squares Regression. The method’s skill is achieved via information on the state of long-term global warming as well as the state and recent evolution of the El Niño-Southern Oscillation and the Interdecadal Pacific Oscillation. We test the out-of-sample skill of the methodology using a “forecast mode” where statistical predictions are made precisely as they would have been if the procedure had been operationalized starting in the year 2000. The forecast errors for lead times of 1 to 4 years are smaller than naïve benchmarks using persistence and perform favorably relative to most dynamical Global Climate Models retrospectively initialized to the observed state of the climate system. Thus, this method can used as a computationally-efficient benchmark for dynamical model forecast systems.