IPCC Report.pdf - Adam Curry

Changes in Climate Extremes and their Impacts on the Natural Physical EnvironmentChapter 3that are known to modulate storm behavior (e.g., ENSO and MJO), isimportant for context when interpreting model output (Sections 3.2.3.2and 3.4.2).While detection of long-term past increases in tropical cyclone activityis complicated by data quality and signal-to-noise issues (as statedabove), theory (Emanuel, 1987) and idealized dynamical models(Knutson and Tuleya, 2004) both predict increases in tropical cycloneintensity under greenhouse warming. Recent simulations with highresolutiondynamical models (Oouchi et al., 2006; Bengtsson et al., 2007;Gualdi et al., 2008; Knutson et al., 2008; Sugi et al., 2009; Bender et al.,2010) and statistical-dynamical models (Emanuel, 2007) consistentlyfind that greenhouse warming causes tropical cyclone intensity to shifttoward stronger storms by the end of the 21st century (2 to 11% increasein mean maximum wind speed globally). These and other models alsoconsistently project little change or a reduction in overall tropicalcyclone frequency (e.g., Gualdi et al., 2008; Sugi et al., 2009; Murakamiet al., 2011), but with an accompanying substantial fractional increasein the frequency of the strongest storms and increased precipitationrates (in the models for which these metrics were examined). Currentmodels project changes in overall global frequency ranging from adecrease of 6 to 34% by the late 21st century (Knutson et al., 2010). Thedownscaling experiments of Bender et al. (2010) – which use an 18-model ensemble-mean of CMIP3 simulations to nudge a high-resolutiondynamical model (Knutson et al., 2008) that is then used to initialize avery high-resolution dynamical model – project a 28% reduction in theoverall frequency of Atlantic storms and an 80% increase in the frequencyof Saffir-Simpson category 4 and 5 Atlantic hurricanes over the next 80years (A1B scenario).The projected decreases in global tropical cyclone frequency may be dueto increases in vertical wind shear (Vecchi and Soden, 2007c; Zhao etal., 2009; Bender et al., 2010), a weakening of the tropical circulation(Sugi et al., 2002; Bengtsson et al., 2007) associated with a decrease inthe upward mass flux accompanying deep convection (Held and Soden,2006), or an increase in the saturation deficit of the middle troposphere(Emanuel et al., 2008). For individual basins, there is much moreuncertainty in projections of tropical cyclone frequency, with changes ofup to ±50% or more projected by various models (Knutson et al., 2010).When projected SST changes are considered in the absence of projectedradiative forcing changes, Northern Hemisphere tropical cyclone frequencyhas been found to increase (Wehner et al., 2010), which is congruentwith the hypothesis that SST changes alone do not capture the relevantphysical mechanisms controlling tropical cyclogenesis (e.g., Emanuel,2010).As noted above, observed changes in rainfall associated with tropicalcyclones have not been clearly established. However, as water vapor inthe tropics increases (Trenberth et al., 2005) there is an expectation forincreased heavy rainfall associated with tropical cyclones in response toassociated moisture convergence increases (Held and Soden, 2006). Thisincrease is expected to be compounded by increases in intensity asdynamical convergence under the storm is enhanced. Models in whichtropical cyclone precipitation rates have been examined are highlyconsistent in projecting increased rainfall within the area near thetropical cyclone center under 21st century warming, with increases of3 to 37% (Knutson et al., 2010). Typical projected increases are near20% within 100 km of storm centers.Another type of projection that is sometimes inferred from the literatureis based on extrapolation of an observed statistical relationship (seealso Section 3.2.3). These relationships are typically constructed on pastobserved variability that represents a convolution of anthropogenicallyforced variability and natural variability across a broad range of timescales. In general, however, these relationships cannot be expected torepresent all of the relevant physics that control the phenomena ofinterest, and their extrapolation beyond the range of the observedvariability they are built on is not reliable. As an example, there is astrong observed correlation between local SST and tropical cyclonepower dissipation (Emanuel, 2007). If 21st-century SST projections areapplied to this relationship, power dissipation is projected to increase byabout 300% in the next century (Vecchi et al., 2008; Knutson et al.,2010). Alternatively, there is a similarly strong relationship betweenpower dissipation and relative SST, which represents the differencebetween local and tropical-mean SST and has been argued to serve asa proxy for local potential intensity (Vecchi and Soden, 2007a). When21st-century projections of relative SST are considered, this latterrelationship projects almost no change in power dissipation in the nextcentury (Vecchi et al., 2006). Both of these statistical relationships canbe reasonably defended based on physical arguments but it is not clearwhich, if either, is correct (Ramsay and Sobel, 2011).When simulating 21st-century warming under the A1B emission scenario(or a close analog), the present models and downscaling techniques as awhole are consistent in projecting (1) decreases or no change in tropicalcyclone frequency, (2) increases in intensity and fractional increases innumber of most intense storms, and (3) increases in tropical cyclonerelatedrainfall rates. Differences in regional projections lead to lowerconfidence in basin-specific projections of intensity and rainfall, andconfidence is particularly low for projections of frequency withinindividual basins. More specifically, while projections under 21st-centurygreenhouse warming indicate that it is likely that the global frequencyof tropical cyclones will either decrease or remain essentially unchanged,an increase in mean tropical cyclone maximum wind speed is also likely,although increases may not occur in all tropical regions. This assessmentis essentially identical with that of the recent WMO assessment (Knutsonet al., 2010). Furthermore, while it is likely that overall global frequencywill either decrease or remain essentially unchanged, it is more likelythan not that the frequency of the most intense storms (e.g., Saffir-Simpson category 4 and 5) will increase substantially in some oceanbasins, again agreeing with the recent WMO assessment (Knutson et al.,2010). Based on the level of consistency among models, and physicalreasoning, it is likely that tropical cyclone-related rainfall rates willincrease with greenhouse warming. Confidence in future projections forparticular ocean basins is undermined by the inability of global modelsto reproduce accurate details at scales relevant to tropical cyclone162