| 英文摘要: | Many tropical countries are exceptionally vulnerable to changes in rainfall patterns, with floods or droughts often severely affecting human life and health, food and water supplies, ecosystems and infrastructure1. There is widespread disagreement among climate model projections of how and where rainfall will change over tropical land at the regional scales relevant to impacts2, 3, 4, with different models predicting the position of current tropical wet and dry regions to shift in different ways5, 6. Here we show that despite uncertainty in the location of future rainfall shifts, climate models consistently project that large rainfall changes will occur for a considerable proportion of tropical land over the twenty-first century. The area of semi-arid land affected by large changes under a higher emissions scenario is likely to be greater than during even the most extreme regional wet or dry periods of the twentieth century, such as the Sahel drought of the late 1960s to 1990s. Substantial changes are projected to occur by mid-century—earlier than previously expected2, 7—and to intensify in line with global temperature rise. Therefore, current climate projections contain quantitative, decision-relevant information on future regional rainfall changes, particularly with regard to climate change mitigation policy. Climate change is expected to drive changes in tropical rainfall by affecting both atmospheric moisture and circulation8, 9, 10, 11. In the absence of circulation change, the enhanced capacity of warmer air to contain moisture would lead to increased P − E (precipitation minus evaporation) in already wet regions and decreases in dry regions; the so-called ‘wet-get-wetter, dry-get-drier’ hypothesis8, 9, 11. This mode of change is present in both climate model simulations9, 11, 12, 13 and observed trends13, 14, and hence is an important validation of model fidelity, but is seen only when very large area averages are used5, 6. At the regional scales more relevant to climate change impacts, the wet-get-wetter, dry-get-drier paradigm is not a good predictor of rainfall change in projections5, 6, 15 or observations16, 17, and projections of future regional rainfall change vary widely across climate models2, 3, 18, 19. Instead, the dominant driver of regional rainfall change in the tropics is the occurrence of shifts in the position of wet regions5, 6. These spatial shifts can cause both increases and decreases in rainfall, and are illustrated here with two very different climate model projections of future precipitation (Fig. 1a, b). Large shifts occur in both models, and in each model are generally coherent in sign over areas large enough to affect whole countries or regions, but the locations of shifts differ greatly between the two. Uncertainty over which regions will experience these shifts, and to what extent, is the main cause of spread in regional rainfall projections5, 6, with even the sign of change uncertain in some regions2, 3. This uncertainty impedes planning for adaptation to climate change. For global climate change mitigation policy, however, the precise location of large rainfall changes may be less important than whether or not they will occur, combined with an estimate of their magnitude and areal coverage.
|
