17.1.4.3.2. Precipitation

In general, all AOGCMs simulate only a marginal increase or decrease in annual rainfall. An area-averaged annual mean increase in precipitation of approximately 0.3% for the 2050s and 0.7% for the 2080s over the Pacific Ocean area is projected, either as a consequence of increases in atmospheric concentrations of GHGs or because of the combined influence of GHGs and sulfate aerosols. The projected increase in precipitation is at a maximum during NH summer (June-July-August) for both time periods. A marginal decline in precipitation is projected for the other three regions, particularly during NH summer—suggesting the possibility of reduced water availability (Lal et al., 2001).

17.1.4.3.3. Extreme high temperature and precipitation events

An analysis of model-simulated daily temperature and precipitation data (from CSIRO and ECHAM model experiments) for the present-day atmosphere and for the two future time slices (2050s and 2080s) projects that the frequency of extreme temperatures during the summer is likely to be higher in all four regions. This implies an increased likelihood of thermal stress conditions during the 2050s and more so during the 2080s. Similarly, although the models project a lesser number of annual rainy days, an increase in the daily intensity of precipitation also is projected (Lal et al., 2001). This suggests an increase in the probability of occurrence of more frequent droughts, as well as floods, in the Atlantic Ocean and Caribbean Sea, the Mediterranean Sea, and the Indian and Pacific Oceans.

Figure 17-4: Climate change scenarios for Atlantic Ocean and Caribbean Sea islands as simulated by five AOGCMs for the 2020s, 2050s, and 2080s.

Figure 17-5: Climate change scenarios for Pacific Ocean islands as simulated by five AOGCMs for the 2020s, 2050s, and 2080s.

17.1.4.3.4. ENSO and precipitation variability

Although it is difficult to obtain reliable regional projections of climate change from GCMs, some consistent patterns are beginning to emerge for the Pacific with regard to ENSO and precipitation variability. Although ocean temperatures over most of the western and southern Pacific warm more slowly than the global average, the eastern central Pacific warms faster than the global average in outputs from several AOGCMs (Meehl and Washington, 1996; Timmerman et al., 1997; Knutson and Manabe, 1998; Jones et al., 1999). These results have been broadly interpreted as an El Niño-like pattern (e.g., Cai and Whetton, 2000) because some simulations appear to project more frequent ENSO-like patterns (Timmerman et al., 1997). In the Pacific, climate variability associated with the ENSO phenomenon manifests itself on a seasonal to decadal time scale. This, in turn, affects factors such as temperature, rainfall, wind speed and direction, sea level, and tropical cyclone climatology that are directly associated with numerous climate-related impacts in the region.

Jones et al. (1999) have produced projected ranges of change in average temperature and rainfall for the western Pacific, based on scaled patterns from three independent AOGCMs and a suite of two AOGCMs and a regional circulation model (RCM). These scaled patterns were produced by the regression method described in Hennessy et al. (1998) and Giorgi and Mearns (1999), in which the influences of decadal variability are significantly reduced. Jones et al. (1999) showed that warming in the Pacific region is projected to increase by less than the global average in most cases. Projections of rainfall are constrained by the models' ability to simulate the SPCZ and ITCZ. All models were able to produce aspects of both features, although the eastern part of the ITCZ was not well captured by any of the models; the higher resolution models were the most realistic. The models also did not produce consistent changes to these features, execept for a large increase in rainfall over the central and east-central Pacific. Most of the projected changes across the western Pacific were increases, with significant increases along the equatorial belt from North Polynesia to further east. Possible decreases were noted in some models for Melanesia and South Polynesia in both halves of the year (April-October and November-March).

Climate variability in the Pacific is a combination of seasonal, multi-annual variability associated with the ENSO phenomenon and decadal variability, the latter influencing the ENSO phenomenon itself. The major concern for impacts in the region is not with the mean climate changes described above but with the extremes that are superimposed on those mean changes. Numerous studies describe the likely intensification of rainfall when the mean change ranges from a slight decrease to an increase. For instance, mean decreases of 3.5% over South Polynesia from the mixed-layer GCM (UKHI) produced little change in intensity, whereas an increase of 7.5% over Micronesia halved the return periods of heavy rainfall events (Jones et al., 1999).

GCMs currently project an increase in SSTs of approximately 1°C by the 2050s and increased rainfall intensity in the central equatorial Pacific, which would impact many small island states in that region. Recent variations over the tropical Pacific Ocean and surrounding land areas are related to the fact that since the mid-1970s, warm episodes (El Niño) have been relatively more frequent or persistent than the opposite phase (La Niña). There are indications that the ENSO phenomenon may be the primary mode of climate variability on the 2- to 5-year time scale and that the current large interannual variability in the rainfall associated with ENSO is likely to dominate over any mean effects attributable to global warming (Jones et al., 1999).

Comparisons between observations and model simulations for the Pacific region further indicate that regional warming would be mostly less than the global average because of the large expanse of ocean. However, there has been a strong indication from several model simulations that the least warming would occur in the southern ocean; the greatest warming could be expected in the far west, central, and eastern equatorial Pacific. In the case of rainfall, increases are likely to be greater where warming over the ocean is greatest, although one GCM output showed no increase in rainfall variability between 1960 and 2100. Model variability is likely to be much lower than historical variability because of the great influence of ENSO, particularly in the Pacific (Jones et al., 1999).

17.1.4.3.5. Tropical cyclones

There is no consensus regarding the conclusions of studies related to the behavior of tropical cyclones in a warmer world. Working Group I concludes that current information is insufficient to assess recent trends, and confidence in understanding and models is inadequate to make firm projections (see TAR WGI). Royer et al. (1998), using a downscaled AOGCM coupled with Gray's method for hurricane forecasting, found no significant change in hurricane frequency or geographical extent for the north Atlantic Ocean, the Pacific Ocean, or the Indian Ocean.

Notwithstanding the foregoing conclusions, individual studies have reported the likelihood of a possible increase of approximately 10-20% in intensity of tropical cyclones under enhanced CO₂ conditions (Holland, 1997; Tonkin et al. 1997). This finding is supported by Jones et al. (1999), who conducted an analysis of tropical cyclones from a 140-year simulation of an RCM nested in a coupled AOGCM for the Pacific region (see Box 17-1). Although the preliminary analysis implies that there might be a small decrease in cyclone formation, an increase in system intensity is projected. The pattern of cyclones during phases of ENSO was unchanged, suggesting that the relationship between cyclone distribution and ENSO may continue. The study by Jones et al. (1999) considers that increases in cyclone intensity (10-20%) estimated by Tonkin et al. (1997) and Holland (1997) are highly likely.

IPCC Homepage