METEOROLOGY AND ATMOSPHERIC CHEMISTRY 



73 



TABLE 1 

 Average Minimum and Maximum Temperatures, °C 



Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec. 



Minimum 23 5 23 4 23 6 23.8 23.5 24.0 23.6 23 6 23 8 23 4 23 7 23.8 

 Maximum 30.4 30 5 30.6 31.1 31.4 31.6 318 32.3 32.3 32.1 31.7 30.9 



Extremes of temperature have also been recorded but 

 provide little additional information. The annual minimum 

 value of 21°C has occurred once in the record and the 

 maximum of 34.4°C three times. The values reported by 

 Blumenstock and Rex (1960) are less reliable because of 

 shorter records. As mentioned above, it is possible that the 

 maximum temperatures are overestimated. 



Similarly averaged relative humidity data shown in Fig. 

 lb exhibit a bit more time dependence than the tempera- 

 ture. There is a broad maximum in the early morning fol- 

 lowed by decreasing values as the temp)erature rises in 

 daylight. The afternoon minimum is lower but briefer in 

 the dry season than it is in the wet season. These values 

 are the best estimates available but may be underestimated 

 during the day by several percent if the temperature is 

 overestimated by 0.5°C. The dew point temperature — the 

 temperature at which saturation will occur if the air is 

 cooled — is less sensitive to such error and is a straightfor- 

 ward indicator of the moisture content of the air. The dis- 

 tribution shown in Fig. Ic seems quite different from, but 

 is entirely consistent with, the data in Fig. la and b. The 

 dew p)oint temp)eraturc is strongly dependent upon the 

 time of year, with a broad minimum in the dry season and 

 a broad maximum in the wet season. In both cases there is 

 an increase during the daylight hours. This increase 

 represents a slight increase in the water-vapor mixing ratio, 

 which is consistent with increased evaporation during the 

 day. It is this increase which prevents the relative humidity 

 from decreasing more than it does in the afternoon. 



Precipitation 



Across most of the North Pacific, including the 

 Marshall Islands area, the rainfall increases markedly from 

 mid-latitudes to just north of the Equator. At Wake (19°N) 

 the annual total is 940 mm, at Kwajalein (9°N) about 

 2400 mm, and at Jaluit (6°N) it exceeds 4000 mm. The 

 highest values are in the equatorial trough near 3° to 6°N 

 and the lowest in the subtropical high pressure area at 

 about 25°N, well east of the dateline. Enewetak at 11°N 

 lies near the northern edge of the zone of most rapid 

 decrease of rainfall with latitude. The average rainfall of 

 1470 mm is not distributed uniformly through the year; 

 about 85% comes during the wet season, which starts in 

 April and ends in mid-November. The variability of the 

 rainfall is remarkable, and this factor is a central theme in 

 the discussion which follows. 



There is much contention in the meteorological litera- 

 ture over the applicability of island-based rainfall data to 

 open ocean conditions. This is not an issue here as we are 

 concerned with the rainfall at the atoll, not in the environ- 

 ment in its absence. We note briefly below the limited data 

 available to discern gradients across the atoll. 



The rainfall distribution through the year, which is 

 based on the archived data tabulated by Taylor (1973), is 

 shown in Fig. 2a. The three measures shown for each 

 month are the rainfall amount exceeded in 90, 50, and 

 10% of the years. Thus the amount expected (50% 

 occurrence) in November, about 124 mm, is somewhat 

 less than the 140 ram in the "rainiest December in 10 

 years." It can be seen at once that in certain months there 

 is a very large range of variability. Although the average 

 April has about 40 mm of rainfall, one year in ten may 

 have less than 10 mm; another may have over 260 mm. 

 The record represented here is 32 years long; there are 

 missing data, but 26 to 29 years of monthly values are 

 available in the various months. A longer series would 

 probably not change the annual total much, and because it 

 is inherent in tropical rainfall, the fluctuation evident in 

 Fig. 2a would not be reduced by additional years of obser- 

 vation. 



For this discussion, we let the dry season begin with 

 December. That this is arbitrary can be seen by comparing 

 the 10% value in Fig. 2a for December with the 50% 

 value for November and the 90% value for November with 

 the 10% value for December. When the wet season ends 

 early, the November rainfall is less than 120 mm, whereas 

 when it ends late, the December total exceeds 50 mm. 

 The January, February, and March 50% values all lie 

 between 20 and 35 mm, and the 90% values are uni- 

 formly very small at <10 mm. Many of the dry season rain 

 events are from small cumulus clouds; however, these 

 affect the total amount less than the infrequent distur- 

 bances. We follow the usual designation and let the wet 

 season begin in April despite the small increase in the 50% 

 value; this may be rationalized by the jump in the 10% 

 figure — i.e., some Aprils are very wet. 



The increased rainfalls of May and June are followed 

 by 50% values between 175 and 225 mm in July through 

 September. The maximum is in October, which also has 

 the highest average and the greatest 90% total. November 

 is a transitional month. The number of days with measur- 

 able (>0.25 mm) rainfall is greatest in August at 21 on the 

 average; this figure varies between 10 and 21 for the wet 

 season months, while it is 10 to 15 during the dry season. 



