the middle of the track. CZCS-derived values of the water-leaving radi- 

 ances were within 15 percent of those measured at the ship station (near 

 the center of the track) at the time of the satellite overpass. 



An example of pigment retrieval on the shelf at somewhat higher concentra- 

 tions is provided in Figure 2-3. This track from the Gulf of Mexico is 

 identical to that presented in Gordon et al. (1980), but was reprocessed 

 using the current algorithms. It demonstrates that accuracies of +30 

 percent can be obtained with pigment concentrations as high as 5 y g a " . 

 Further improvements in the pigment retrieval accuracy over this range 

 (~0-5 mg/m ) are unlikely considering that the inherent error in the bio- 

 optical algorithms for Case 1 waters appears to be about +30 percent. 



Emphasis in algorithm development is now being placed on retrieval tech- 

 niques for near-Case 2 waters. Preliminary retrievals in the Mississippi 

 Delta have been within a factor of +2 using the algorithms described above. 

 Considering their definition, one cannot foresee the existence of a 

 universal Case 2 bio-optical algorithm, since phytoplankton do not domi- 

 nate the optical properties of such waters. Instead, site-specific 

 algorithms are to be expected. This poses no problem for MAREX since 

 measurements from buoys are an integral part of the proposed program, and 

 such measurements can be used to provide data to "update" the remotely 

 sensed pigment concentrations if necessary. Also the assumption used in 

 the atmospheric correction algorithm (see Appendix B) that the ocean is 

 black at 670 nanometers is not valid in areas of high sediment suspension, 

 indicating that another spectral band at longer wavelengths in the near 

 infrared should be added to a follow-on CZCS. 



Furthermore, in areas of little river runoff, Smith and Baker (1981) have 

 presented a preliminary time series of CZCS imagery, along with a quantita- 

 tive statistical analysis of each image, of the Southern California Bight, 

 which clearly provide a synoptic view of this complex upwelling region. 

 The sequence of qualitative images reveals significant temporal and spa- 

 tial variation and a richness of detail which is impossible to obtain from 

 shipboard data alone. These chlorophyll maps of the California upwelling 

 regime from the Nimbus-7 CZCS were further used to develop an algorithm for 

 estimating primary production from the satellite imagery (Smith, Eppley 



2-8 



