Other reports in this collection
It is uncertain how a given emissions path converts into atmospheric concentrations of the various radiatively active gases or aerosols. This is because of uncertainties in processes relating to the carbon cycle, to atmospheric trace gas chemistry and to aerosol physics (see Chapters 3, 4 and 5). For these uncertainties to be reflected in climate scenarios that rely solely on GCM outputs, AOGCMs that explicitly simulate the various gas cycles and aerosol physics are needed. At present, however, they are seldom, if ever, represented in climate scenarios.
Even when presented with a given greenhouse gas concentration scenario, there are considerable uncertainties in the radiative forcing changes, especially aerosol forcing, associated with changes in atmospheric concentrations. These uncertainties are discussed in Chapters 5 and 6, but again usually remain unrep-resented in climate scenarios.
An additional set of modelling uncertainties is introduced into climate scenarios through differences in the global and regional climate responses simulated by different AOGCMs for the same forcing. Different models have different climate sensitivities (see Chapter 9, Section 9.3.4.1), and this remains a key source of uncertainty for climate scenario construction. Also important is the fact that different GCMs yield different regional climate change patterns, even for similar magnitudes of global warming (see Chapter 10). Furthermore, each AOGCM simulation includes not only the response (i.e., the “signal”) to a specified forcing, but also an unpredictable component (i.e., the “noise”) that is due to internal climate variability. This latter may itself be an imperfect replica of true climate variability (see Chapter 8). A fourth source of uncertainty concerns important processes that are missing from most model simulations. For instance AOGCM-based climate scenarios do not usually allow for the effect on climate of future land use and land cover change (which is itself, in part, climatically induced). Although the first two sources of model uncertainty - different climate sensitivities and regional climate change patterns - are usually represented in climate scenarios, it is less common for the third and fourth sources of uncertainty - the variable signal-to-noise ratio and incomplete description of key processes and feedbacks - to be effectively treated.
Most climate scenario construction methods combine model-based estimates of climate change with observed climate data (Section 13.3). Further uncertainties are therefore introduced into a climate scenario because observed data sets seldom capture the full range of natural decadal-scale climate variability, because of errors in gridded regional or global baseline climate data sets, and because different methods are used to combine model and observed climate data. These uncertainties relating to the use of observed climate data are usually ignored in climate scenarios. Furthermore, regionalisation techniques that make use of information from AOGCM and RCM experiments to enhance spatial and temporal scales introduce additional uncertainties into regional climate scenarios (their various advantages and diasdvantages are assessed in Chapter 10 and in Section 13.4). These uncertainties could be quantified by employing a range of regionalisation techniques, but this is rarely done.
|
Other reports in this collection |
|