| 英文摘要: | Long-term climate change and sea-level rise in model projections have been primarily determined by external forcing of climate conditions. Now, research shows that centennial projections of the dynamic sea level are also sensitive to the ocean's initial conditions. Unlike in numerical weather forecasting and seasonal to decadal climate prediction, climate modellers have mostly disregarded atmospheric and oceanic initial conditions when performing long-term climate and sea-level projections. Instead, model experiments with changing external forcing are typically branched from random climate states in the unforced control simulation. This is because internal variability, which is sensitive to initial conditions, has been thought not to alter the long-term trend or statistics in forced experiments, just as weather fluctuation during spring does not influence the gradual warming towards summer. But this assumption may need to be modified. Writing in Nature Climate Change, Bordbar and colleagues1 demonstrate that, in centennial projections of the dynamic sea level (DSL), long-term internal variability plays as great a part as external forcing. The DSL is an important, sometimes dominant, factor influencing regional and local sea-level rise. Accurate satellite measurements reveal that the ocean surface is not flat, but shows 'mountains' and 'valleys' with peak-to-trough magnitudes of up to 3 metres. This ocean topography, referred to as the DSL, provides the pressure gradient force that drives large-scale ocean circulation. It is also closely related to ocean temperature, salinity and mass distribution, and can readily alter in response to climate variability and change. Three-dimensional complex models are powerful tools for making projections, including for the DSL, but they come with inherent uncertainty. A key task for climate scientists is to not only identify signals of externally forced changes, but also to quantify and reduce the associated uncertainty. For long-term projections, uncertainty can arise from three distinct sources: internal variability, model imperfection and forcing (scenario) uncertainty2. The relative importance of these sources changes with the temporal and spatial scales under consideration (Fig. 1). For example, internal variability could contribute significantly to near-term and regional prediction of surface air temperature2, but its impact decreases towards longer and larger scales, so that on centennial and global scales, its role is minor and can usually be disregarded.
|
