178
reads

Mike Wallace

2012-09-17
09:15:00 - 10:05:00

101 , Mathematics Research Center Building (ori. New Math. Bldg.)

Many questions concerning the nature and causes of climate variability on the interdecadal time scale are still unresolved. For example, there is no consensus within the scientific community as to whether time-varying forcing associated with aerosols or whether variations in the Atlantic Meridional Overturning Circulation were responsible for the mid-20th century hiatus and the recent (post-1998) slowdown in the rate of global warming. Nor is it clear why the Arctic has experienced rapid warming during the past decade while surface air temperatures over the Northern Hemisphere as a whole have warmed very little or why, during the late 20th century; wintertime temperatures over the Northern Hemisphere continents poleward of 40°N warmed three times as rapidly as global-mean (land plus ocean) annual-mean surface air temperature during the same interval. These spatial and temporal differences in the rate of warming stem from the fact that the climate system is varying on the interdecadal time scale in response to its own internal variability as well as to a variety of natural and anthropogenic forcings. Detection, attribution and projection of human-induced climate change on the decade-to-decade time scale is problematical, and in some cases it is even difficult to distinguish between low frequency climate variability and climate change. These ambiguities can be expected to persist until the signature of human-induced climate change becomes large enough to stand out clearly above the natural “background variability”, as is projected to occur in the second half of this century.

In the world of models, the distinction between climate change and climate variability is clear. Climate change in response to the buildup of greenhouse gases and other anthropogenic forcings can be determined from a suite of simulations in which each member is started from a different set of initial conditions and run with the same prescribed, time-varying external forcings. Provided that the number of individual realizations is large enough to ensure a high level of statistical significance, the ensemble-mean climate can be identified with the human-induced climate change “signal” and the departures of the trajectories in the individual realizations from the ensemble mean trajectory are attributable to the internal variability of the simulated climate system. The same ensemble runs can be used to generate probabilistic projections of climate change over the next few decades.

Just how applicable the results derived from the model world are to the real world depends upon how well the models are able to simulate the internally generated low frequency variability of the climate system. With only one observed climate trajectory that can be used as a basis for validating the models, this question cannot be answered definitively, but it can be assessed in a probabilistic way. Generating a robust characterization of the low frequency variability in the historical climate record with which the statistics derived from the model runs can be compared is a prerequisite for assessing the internally generated variability. Just what such a characterization should include has not yet been fully agreed upon in the climate diagnostics community.

It seems reasonable that a robust characterization of the variability in the historical record should include variance and covariance statistics. Given only one realization, such statistics must be based on temporal, as opposed to run-to-run variances and covariances. To obtain a robust characterization of the variability (i.e., a characterization with a sufficient number of statistical degrees of freedom to be considered reliable) it is necessary to restrict the analysis to frequencies 10-30 times higher than one cycle over the length of the historical record , e.g., through the application of a high pass filter. Attribution of variability with frequencies lower than this cutoff frequency is inherently ambiguous. It is amenable to probabilistic statements based on comparisons with statistics derived from ensembles of runs, but there is no way of knowing with certainty whether it is natural or human-induced. Given that the historical record of global observations of surface air temperature and precipitation is only a ~100-years long, it is inevitable that decadal variability falls within this “twilight zone” in which attribution can be performed only in a probabilistic way.

Another factor that limits our ability to diagnose the decadal-scale variability in the climate record is the fact that inherently stochastic variability on the interannual time scale associated, for example, with the ENSO cycle or with large excursions of the Northern and Southern Hemisphere annular modes is capable of inducing a substantial sampling variability on the interdecadal time scale. For example, it has been questioned whether the so-called Pacific Decadal Oscillation (PDO) is merely a manifestation of such stochastic, sampling variability. It has also been suggested that anthropogenic forcing could also be responsible for some of the observed circulation changes on the multidecadal time scale. For example, the more rapid warming of the continental interiors relative to the surrounding oceans due to the much lower heat capacity of the underlying surface could serve to weaken the wintertime planetary waves .

Regardless of the mechanisms that give rise to it, decadal scale climate variability mediates the rate of rise of global-mean temperature. Performing a “dynamical adjustment” to remove (or at least reduce) the contribution of these circulation changes to the rise in global-mean temperature simplifies the space-time structure of the surface air temperature record and renders it more spatially and seasonally coherent.

In this presentation we will summarize the state of our knowledge of internally generated interdecadal variability of the climate system. We will show how the interplay free and forced climate variability complicates the attribution of global warming, regional climate impacts and extreme events and we will demonstrate how performing a dynamical adjustment can simplify the representation of climate change in the historical record.

For material related to this talk, click here.