The regression equation computed from the 

 Laevastu formulation yields the lowest stand- 

 ard error of estimate, but it is immediately 

 apparent that the Laevastu equation grossly 

 overestimates the daily total as judged by the 

 mean value of the 104 test observations, as 

 well as the regression equation (table 2). As 

 was true with the other equations, changing 

 the constants in the equations could improve 

 agreement between the observed and predicted 

 values. The tendency of this equation to over- 

 estinnate incident solar radiation under cloud- 

 less skies is also apparent in a comparison of 

 the tabulated values of Budyko (1956) at 0, 10, 

 and 20°N. for the 15th day of eachmonth of the 

 year with that computed by the Laevastu 

 equation. The total of incident solar radiation 

 for the 36 days was 22,941 g. cal. per square 

 centimeter according to Budyko (1956; table 2) 

 and 26,887 g. cal. per square centimeter ac- 

 cording to the Laevastu equation. The daily 

 difference amounts to 101 g. cal. per square 

 centimeter. The results of this comparison and 

 the one above suggest that the Laevastu equa- 

 tion has a systematic error which should be 

 evaluated before application in tropical ocean 

 areas. 



It seems doubtful that equations such as 

 those discussed above will provide adequate 

 estimates of daily insolation on any given date 

 at any location in the ocean. Direct measure - 

 nnent of incident solar radiation is certainly 

 to be desired, although more complicated 

 equations that use data on relative humidity, 

 and type, amount, and height of clouds will 

 probably give better estimates than do the 

 simpler equations. 



SUBMARINE IRRADIANCE 

 Definitions 



Submarine irradiance was originally meas- 

 ured to determine the rate of attenuation of 

 diffuse downwelling irradiance with depth and 

 its relation to photosynthetic rates. During this 

 study it became evident that the average dif- 

 fuse attenuation coefficient, k, was significantly 

 correlated with chlorophyll a, and, in fact, 

 could be successfully substituted on occasion 

 for chlorophyll a in certain statistical tech- 

 niques that lead to the prediction of zooplankton 

 standing crop. Thus the irradiance measure- 

 ments obtained with the equipment described 

 below have been used in two separate contexts, 

 which I describe elsewhere. 



Irradiance at a specified wavelength is de- 

 fined as follows: 



H = I N cos 9 duj 



■'o 



where H is radiance, 6 is the angle of incidence 

 and o) the solid angle measure (Preisendorfer, 



1960). The value of the diffuse attenuation 

 coefficient k is given by 



H, 



Hv 



-kiZ 



or, in the more common form 



loge ^zi - loge Hz 2 



Z2 - Zi 



where ^z^ and Hzp are irradiance values at 

 the shallower and deeper depths, respectively. 

 H must be specified as to direction (down- 

 welling or upwelling) and wavelength. 



In this paper k is always calculated from 

 downwelling irradiance values from the third 

 equation given above. The assignment of wave- 

 length specificity is discussed below. 



Instrumentation 



The underwater and deck detector units 

 contained a Weston self-generating barrier 

 layer cell (Type 856RR) mounted in watertight 

 housings. Optical coupling between the water 

 and the barrier layer cell was obtained with 

 an abraded translucent plastic collector with 

 cosine collecting properties. The deck unit 

 was mounted in gimbals above most of the 

 vessel's superstructure, to give a nearly 

 uninterrupted view of the sky. 



A Wratten No. 45 filter (blue-green) was used 

 in the underwater unit. The half-band width of 

 the complete unit is estimated to be about 63 

 mn and to have peak transmittance centered at 

 490 m^ (table 4). In ocean water the trans- 

 mittance peak and half -band width of the unit 

 changes with depth as a result of the inter- 

 action between the changing spectral distri- 

 bution of the water and the detector's nonlinear 

 spectral response. 



Most of the measurements in this study were 

 made in water types I, II, or III (Jerlov, 1951: 

 table 9), and calculations show that the attenu- 

 ation coefficients measured with the equipment 

 may be assigned to 475 m/i. 



The Irradiance Meter 



A number of different submarine irradiance 

 meters provided with nearly identical detector 

 units have been used to measure relative 

 downwelling irradiance. Diffuse attenuation 

 coefficients have been calculated from irradi- 

 ance values obtained at different depths in the 

 upper 20 to 1 20 m. 



Each detector unit (fig. 1) consists of a 

 photovoltaic cell housing (D) and a threaded 

 cap (C). The metal or plastic housing (D) con- 

 tains a photovoltaic cell (G) and an optical 

 filter (E). Silica gel, placed in the cavity 



