A Precipitation Climatology for Stations in the Tropical Basin;
Effects of ENSO
To provide an overview of the island climatologies, Fig. 2 shows the annual cycle (over 12 running 3-month periods) of the climatological mean and the year-to-year variability of the seasonal total rainfall at all 66 Pacific Island stations, based on the period 1955-96. Shown by box-and-whiskers plots are the median, the 25th and 75th percentiles, and the extreme minimum and maximum amounts for the given station/season. A tendency for relatively wet boreal (Northern Hemisphere) winters and dry summers is clear at the four Hawaiian stations (but only weakly at Hilo). However, this pattern is more the exception than the rule in the Tropics, as relative wetness in late local summer to fall (July to October in the northern Tropics, January to April in southern Tropics) is apparent at most of the other islands. Many of the stations in Papua New Guinea and northeastern Australia have especially marked seasonal cycles with very little rain during the local winter seasons. Because equatorial Fanning and Christmas islands are surrounded by a somewhat cool ocean most of the year, they receive fairly light climatological rainfall, but with very large positive deviations occurring during El Niño episodes. Many of the more off-equator locations, especially west of 170.5W, experience drought with El Niño. Johnston Island has a somewhat hybrid climate, with wetness in fall and early winter, and minimum rainfall in early summer.
The variability of the basin rainfall is further characterized in Table 2 for the four "regular" seasons of the year (Jan-Feb-Mar, Apr-May-Jun, Jul-Aug-Sep, and Oct-Nov-Dec). The seasons are chosen as such because of the lag of the sea-surface temperature (SST) relative to the solar seasonal cycle. In the northern Tropics, for example, a maximum in SST (and in the associated warm, moist island climate) tends to occur during Jul-Aug-Sep rather than in Jun-Jul-Aug which is more traditionally regarded as summer in continental climates. Table 2 indicates for each station the median rainfall amount (in parentheses in the top row), and the percentages of the median corresponding to decile boundaries ranging from the 10 %ile to the 90 %ile. The extreme cases, representing the highest and lowest recorded percentages of the median, are also indicated. Because there are 42 cases for each station/season, the highest and lowest correspond to the 98.8 and 1.2 %iles, respectively. (This uses the center of a 1/42, i.e. the midpoint of a 2.4 percent, interval [yielding 1.2 percent] to represent the ranks of 1 and 42 for the highest and lowest cases, respectively.) It is noted that stations that have wet climatologies (e.g. Pohnpei in most seasons) tend to have narrow distributions in terms of percentages of median amount, while the normally drier stations (e.g. Honolulu, Christmas Island) more frequently take on comparatively extreme ratios with respect to the median. In the case of Honolulu, the implication for water supply may not be critical in view of its leeward position relative to the local topography, and the accessibility of water from nearby windward watersheds that receive more plentiful rainfall. This points up the need for a more detailed analysis and understanding of the spatial variation of rainfall on islands that have substantial topography.