Previous: Introduction Next: Summary Up: Ext. Abst.

The elements of detection and attribution of climate change

Here, I briefly discuss the essential elements used in the detection and attribution of climate change, with an emphasis on the role of the stratosphere. Anthropogenic climate change, if and when itf exists, must occur against a background of natural internal climate variability and natural forced climate variations. Detection and attribution of climate change is a statistical ‘signal in noise’ problem. Detection is the process of demonstrating that an observed change is significantly larger than could be explained by natural internal variability. Attribution is the process of demonstrating that the observed changes are consistent with the climate response to a specified forcing and not consistent with the response to other forcings. Hence, any detection and attribution study needs to consider observational evidence of climate change, estimates of internal climate variations, estimates of natural and anthropogenic forcing factors and the climatic response to such forcings as simulated by climate models, as well as quantitative methods to compare observed climate changes with the simulated responses to different forcings.

The longest instrumental climate records exist for surface air temperatures, with data available for some regions since about 1850 and adequate coverage to provide a reliable estimate of global-mean temperature since about 1900. These data show a warming of global-mean temperature by about 0.6± 0.2° C over the last 100 years. Data for the stratosphere is available for a much shorter period, with radiosonde data available in the lower stratosphere since about 1960 and satellite data available from the 1970s (NASChanin and Ramaswamy, 1999). These data show a global-mean cooling in the lower stratosphere of about 0.2° C per decade, interrupted by irregular warming period associated with major volcanic eruptions. Higher in the stratosphere, the cooling trends are even larger (Chanin and Ramaswamy, 1999)

In order to assess whether the changes in surface or lower stratospheric temperatures noted above are due to external climate forcing factors or to internal climate variability, reliable estimates of the magnitudes and patterns of internal climate variations are required. The observational data for both the surface and the lower stratosphere are too short to be used to estimate internal climate variability. In addition, they may have been affected by the response to external climate forcings, making it difficult to isolate the effects of internal variability alone. The only way to estimate internal climate variability is to use very long simulations of climate models that have no variations in external forcings. Such control simulations have been run with a number of different coupled ocean-atmosphere climate models for periods of 1000 years or longer. They show that the observed warming over the last 100 years is very unlikely to be due to internal climate variability alone.

However, such models do not provide reliable estimates of internal climate variability in the stratosphere, as they have very poor stratospheric resolution (Gillett et al, 2000). Unfortunately, no climate models with adequate resolution in the stratosphere have been run out for extended periods of several hundred years, which would be needed to estimate internal climate variability on multi-decadal time scales. Simulations with some simpler stratospheric models show large multi-decadal variability in the stratosphere due to internal dynamics alone. In addition, it is likely that the coupling between atmospheric circulation and chemistry plays a role in the internal variability of the stratosphere. No long control simulations are available from climate models with adequate stratospheric resolution and interactive chemistry. Hence, estimates of internal climate variability in the stratosphere are uncertain.

If significant climate changes can be detected in the observational data, the next step is to compare them with the climate responses to different forcings to identify the possible causes, both natural and anthropogenic. Reliable estimates exist for the magnitude and patterns of changes in anthropogenic forcing due to increasing greenhouse gases, but there is some uncertainty in the forcing due to stratospheric ozone changes and greater uncertainty in the forcing due to changing stratospheric ozone or due to changing tropospheric aerosols due to human activity. In addition to these anthropogenic forcings, changes in natural external climate forcings have been included in this IPCC assessment. Estimates of the climate forcing due to changing solar irradiance or due to changes in stratospheric volcanic aerosols are based on direct observations for the last two decades but are based on indirect evidence over the last 100-200 years and therefore have greater uncertainty. The response to these forcings must be estimated from simulations with climate models. The response to individual major volcanic eruptions seems to be well-simulated but changes in volcanism appears to have played a small role in recent climate variations. The response to the direct effect of changes in solar irradiance has been simulated in climate models. However, those simulations used in detection and attribution studies have not included possible changes in stratospheric ozone associated with changes in solar UV irradiance. This coupling between ozone and climate is likely to be important in simulating the climate response to changes in solar irradiance. Another climate forcing in the stratosphere which has received little attention until recently is changing stratospheric water vapour. While there is some evidence of trends in water vapour in the upper troposphere and lower stratosphere (SPARC WAVAS, I don’t have the correct reference), this observational evidence is uncertain and the climate response even less certain.

A number of different approaches have been taken for the quantitative comparison of observed and modelled climate changes due to different causes. Early studies compared the magnitude of variations of global mean temperature between observations and climate models with anthropogenic forcings. Later, fingerprint studies compared the spatial patterns of observed temperature changes with those forced by natural and anthropogenic forcing factors. For example, Tett et al (1996) showed that the inclusion of increasing greenhouse gases, decreasing stratospheric ozone and increasing tropospheric sulphate aerosols were needed so that model simulations of the zonal mean temperature variations in the troposphere and lower stratosphere over the last four decades compared well with observed temperature variations. Most recently, optimal fingerprint studies have compared the spatial and temporal variations of the observed climate with model simulations with different forcings. These show that the observed warming over the last 40 years cannot be explained by natural forcing factors alone, such as changes in solar irradiance or volcanic activity, as the observed changes in these forcings would be expected to lead to a cooling over this period (Tett et al, 1999). It is only through the inclusion of increasing greenhouse gases, increasing tropospheric aerosols, and decreasing stratospheric ozone as forcings in climate model that they provide simulations consistent with the large-scale observed temperature changes over the last 40 years.