74 



MERRILL AND DUCE 



WET SEASON 

 DRY SEASON 



(b) 



RAINFALL DEVIATION 



Fig. 2 Rainfall data for Enewetak. (a) Rainfall amounts for 

 each month. The three measures correspond to the amount, 

 in mm, exceeded in 90% (below shading). 50% (above) and 

 10% (top) of the months in the record, (b) Deviation frequen- 

 cies for the wet and dry seasons. The curve shows the percent 

 occurrence of deviations from the seasonal mean in units of 

 a. the square root of the variance. The corresponding rain 

 amounts, in mm are shown below the axis. 



Thus, as noted above, there are many small rain showers 

 even in the dry season. The wet season amount includes 

 such cumulus showers and a greater number of larger, sus- 

 tained rainfalls during disturbances and during the infre- 

 quent tropical storm or approach of the equatorial trough; 

 these are much less common in the dry season. 



Another way to look at the variability of the precipita- 

 tion is shown in Fig. 2b, where the frequency of 

 occurrence of monthly amounts is given as a function of 

 the deviation from the season average. Note first that in 

 both regimes the most common value is significantly lower 

 than the mean, about 0.5 a below. (Here a is the root 

 mean square deviation of rainfall amounts for the season.) 



In the dry season, very dry months are common, but a few 

 months with large amounts of rainfall do occur. For exam- 

 ple, 2% of the years would be expected to have a "dry" 

 month with 160 mm of rain, 2.5 a above the average. 

 Note also that the overall occurrence of the large rainfall 

 months is not dependent upon the season but that the 

 amount of precipitation is. The infrequent 3 or 4 a cases 

 for the wet season corresfsond to very substantial totals. 

 These frequency distributions are typical of subtropical 

 sites but are somewhat uncertain far from the mean value 

 because of the limited length of record. Also, since these 

 frequencies were averaged over the entire season, the 

 month-to-month variation, which is large in the wet season, 

 is lumped together with the interseasonal difference here. 



Insufficiency of data limits one's ability to document 

 diurnal variation In precipitation amount. Nevertheless, 

 Lavoie (1963), using primarily Enewetak data, presented 

 convincing evidence of an early morning maximum in 

 frequency of rain; the deviation is fjerhaps 15% at the 

 peak. There is some evidence in the same data set for a 

 broad and weak afternoon minimum in the rain 

 occurrence. Lavoie considered several mechanisms in an 

 attempt to rationalize these values and to explain some sig- 

 nificant difficulties: the data have large scatter, and even 

 the maximum does not appear at every station. It thus 

 seems best to say that there is a tendency for a maximum 

 in the rainfall occurrence in the early morning and a weak 

 minimum in the afternoon. 



Even more limited are data giving the spatial distri- 

 bution of rain about the atoll. Any variation is assumed to 

 be primarily random because of the low relief of the 

 atoll — i.e., the absence of orographic forcing. However, 

 there could be sufficient disruption of the thermodynamic 

 structure of the atmosphere by the presence of the lagoon 

 to cause a discernable pattern. Data laboriously collected 

 by Blumenstock and Rex (1960) for six special sites on 

 islands around the atoll during 2-week periods, once in 

 each season, have not to our knowledge been carefully 

 analyzed in the literature. They reveal no systematic pat- 

 tern of variation. The rainfall amounts at the various sta- 

 tions are highly correlated only when the stations are close 

 together, and there is always some difference among them. 

 The record thus appears consistent with rain areas of vari- 

 ous sizes unforced by the atoll itself. Nevertheless, this 

 does not rule out some such forcing in other cir- 

 cumstances. This record does not include any disturbed 

 weather periods during which there could be a measurable 

 difference of rainfall across the atoll. 



Cloud Cover and Solar Radiation 



Accurate estimates of cloud distribution and type are 

 not easy to obtain, particularly at night and when low 

 clouds obscure the sky. As discussed by Blumenstock and 

 Rex (1960), there is likely a systematic bias — 

 overestimation. Fortunately the overall cloud amount is 

 least affected, and voluminous data exist for this quantity 

 in the archive. Again, the average variation with time of 

 day and time of the year is the main topic of discussion. 



