for <C> >lmg/m 3 CR(13)] and one for < C > >lmg/m 3 CR(23)3. A similar 



approach is still being used to process CZCS imagery: the regression 



3 

 line in Figure B-3 being used for <C > >1.5 mb/m , and the regression 



3 

 line in Figure B-5 f or < C > >1.5 mg/m . The rational for these choices 



is that low pigment applications would usually involve mostly Case 1 



waters, while the higher concentrations are likely to be a mixture of 



Case 1 and Case 2 waters. 



All of the linear regressions shown in these figures are of the form 



Log<C(i,j)> = Log A(i,j) + B(i,j) Log R(i,j). (6) 



p 

 The values of A, B, r , the standard error of estimate, s, and the 



number of samples in the regression, N, for the various algorithms are 



presented in Table B-l. The relative error in <C> is approximately 10 



- 1. The specific algorithm now being used by NASA to compute <C> is: 



<C> = <C>, if <C>, <1.5, 



<C> = <C>, if <C>, >1.5 but <C>, <1.5, 



<C> = <C>, if <C>i >1.5 and <C>, >1.5, 



3 

 where <C>. is in mg/m and <C>, and <C> 3 refer to algorithms 1 and 3 in 



Table B-l. 



B.2 ATMOSPHERIC 



To understand the physics of atmospheric correction, consider a physical 

 setting wherein solar irradiance F ( x) at a wavelength x is incident on 

 the top of the atmosphere at a zenith angle e and azimuth <f> and the 

 scanner is detecting radiance L t (x) at a nadir angle e and azimuth $. 

 L.(x) consists of radiance which has been scattered by the atmosphere 

 and sea surface, and radiance generated by Fresnel reflection of the 

 direct (unscattered) solar irradiance from the rough ocean surface (sun 

 glint), as well as radiance which has been backscattered out of the 

 water t(x)L (x). The goal of the atmospheric correction is to estimate 

 t(x)L.(x) on a pixel by pixel basis from measurement of L t (x). This 

 requires removal of the radiance added by interactions with the 



B-9 



