英文摘要: | A key question surrounding proposals for climate engineering by increasing Earth's reflection of sunlight is the feasibility of detecting engineered albedo increases from short-duration experiments or prolonged implementation of solar-radiation management. We show that satellite observations permit detection of large increases, but interannual variability overwhelms the maximum conceivable albedo increases for some schemes. Detection of an abrupt global average albedo increase <0.002 (comparable to a ~0.7 W m−2 reduction in radiative forcing) would be unlikely within a year, given a five-year prior record. A three-month experiment in the equatorial zone (5° N–5° S), a potential target for stratospheric aerosol injection, would need to cause an ~0.03 albedo increase, three times larger than that due to the Mount Pinatubo eruption, to be detected. Detection limits for three-month experiments in 1° (latitude and longitude) regions of the subtropical Pacific, possible targets for cloud brightening, are ~0.2, which is larger than might be expected from some model simulations.
Proposals for deliberate modification of the climate system to counteract anthropogenic climate change are gaining momentum1, 2, 3, 4, 5, 6, 7. Prominent among the proposed technological fixes (collectively dubbed climate engineering or geoengineering) is an assortment of sunlight-reflection methods (SRM, also called solar-radiation management or short-wave climate engineering) that includes: injecting reflective particles into the stratosphere; brightening marine stratocumulus clouds in the troposphere; and increasing the reflectivity of the Earth's surface, including vegetated areas, oceans and built environments. Their common aim is to modify Earth's energy balance to maintain an average surface temperature within some acceptable range by increasing planetary albedo (reflectivity) to reduce absorption of incoming short-wave radiation. A growing body of literature addresses scientific, technical, environmental, ethical and legal issues1, 2, 3, 4, 5, 6, 7. However, it largely ignores the question: could we detect the impacts of either a planned, publicized or independent, undisclosed climate engineering effort? The answer depends on the availability and adequacy of global observations and on the background variability of the climate system. Although several workshop and committee reports1, 2, 4 mention the general problem of detecting engineered albedo changes (and one7 outlines observational requirements for monitoring incoming and reflected solar radiation), no analysis so far has quantitatively estimated detection limits for engineered albedo changes. Studies have considered detection of temperature (and precipitation8) changes that might result from SRM activities using either rough estimates7 or model simulations8 of the background variability. Because changes in these variables have a direct impact on ecosystems and societies, they are of critical interest in SRM discussions. However, temperature and precipitation can change not only in response to changes in climate forcings, but also because of natural climate variability. Thus detection of albedo changes due to SRM is fundamental to determining the effectiveness of SRM in changing Earth's radiative balance and in causing subsequent changes in surface climate variables. This analysis uses observations of incoming and outgoing short-wave radiation to estimate detection limits for hypothetical SRM-induced albedo increases in an effort to help frame discussions of potential SRM field experiments and implementation.
Detection of albedo changes from SRM, like climate change detection in general, is essentially a signal-to-noise problem. The climate system perturbation must exceed both measurement uncertainty and climate variability. Complications arise from the expectation that SRM signals will have spatial and temporal structure. A large regional albedo increase might occur in an area that is not well observed, might be offset by albedo decreases elsewhere or might not appreciably change the global average. A response might not take place immediately, either because of lags in the climate system or because the engineering activity might involve a gradual change. The activity might not cause a sustained effect because the intervention is short term, because climate system processes damp the initial local signal9 or because the system tends to rebalance to maintain a stable global or hemispheric value10. One can foresee such signal-detection complications arising with the leading SRM proposals: - marine stratocumulus cloud-brightening effects in one region causing opposite albedo effects nearby due to associated changes in atmospheric dynamics9
- aerosol injection into the stratosphere causing a gradual increase in albedo as concentrations increase and a gradual decrease as particles leave the atmosphere
- SRM being attempted in a region lacking long-term observations
- governance mechanisms endorsing SRM experiments whose proposed impact is below some regulatory threshold11 or which are anticipated to be measurable locally but not environmentally significant at larger scales12
This analysis used standard statistical tests and existing observations13, 14, 15 to estimate detection limits for idealized albedo interventions. We considered detection at both the global and regional scales, distinguished between abrupt, sustained interventions and gradual ramp-up schemes, and assessed how the duration of an experiment influences its detection. Because we used existing data, the analysis could not investigate the effects of potential feedbacks induced by the perturbation. Details regarding data and test procedures are in the Methods section. Uninterrupted, near-global, high-precision records of incoming and reflected solar radiation measured by satellite-borne instruments are required for detecting SRM activities. Whether the intended albedo increase is at the surface (for example, light versus dark coloured roofs), in the troposphere (cloud brightening) or in the stratosphere (aerosol injection), the goal is to increase reflection from the planet, so space-based measurements are well-suited to the detection problem. We employed albedo computed from the 2000–2012 CERES EBAF Ed2.6r13, 14, 15 observations to simulate SRM interventions and to determine how large an intervention must be for it to be detected above the variability of the climate system. The monthly 1° latitude × 1° longitude (~1010 m2) resolution of the data frames the space and time resolution of both this analysis and potential future SRM detection based on these observations.
The annual cycle causes the most salient variations in albedo16 (Fig. 1a–d). At high latitudes, snow and ice accumulation in winter produce the largest seasonal changes (~0.4 non-dimensional albedo units), but at lower latitudes, changes in vegetation and cloudiness also cause significant albedo changes (~0.2). The amplitude of the annual cycle in global average albedo is 0.03, ~10% of the annual mean value (0.29); see Methods and Fig. 2. We examined two areas of the eastern subtropical Pacific Ocean, each at both 1° × 1° and 5° × 5° resolution, where the annual cycle is a much larger percentage of the annual mean, and a 1° × 1° area of the central equatorial Pacific, where both mean albedo and albedo variability are very low (Fig. 1 and Methods). The amplitude of the annual cycle is ~0.10 and 0.25, in the northern and southern subtropical locations (NP and SP in Fig. 1), respectively, or ~37% and 78% of the annual mean values of 0.27 and 0.32. In the equatorial location (CP in Fig. 1), the annual mean albedo is 0.13, and the amplitude of the annual cycle is ~6% of the mean.
|