Data Files in this set: sbmer*.cdf - these are vertical casts including Lu, Ed, Es, etc., extrapolated to the surface (0-) and binned at 1 m intervals. These files can be read with read_sbmer.c. The Lu data in *.sbmer files has not been corrected for the instrument self-shading. Self-shading correction is being done only for the surface-extrapolated values in a separate file. The difference is usually less than 3% (Kahru and Mitchell, 1998. Evaluation of instrument self-shading and environmental errors on ocean color algorithms, Proceedings of Ocean Optics XIV, CD-ROM, Kona, Hawaii, eds. S. Ackelson & J. Campbell, Nov 1998. In most stations more than 1 vertical casts were made, however, only one cast (the best) per station is included here. ac*.cdf - these are vertical casts including 8 wavelengths of agp and 9 wavelengths of c obtained with the AC9 Absorption and Attenuation meter. Depth-binned at 1 m intervals. These files can be read with read_ac.c. H.cdf - contains vertical casts including 6 wavelengths of bb obtained with the Hydroscat-6 backscattering meter. Depth-binned at 1 m intervals. This file can be read with read_H.c. chldepths_jes9906.cdf - has all the depth samples of CHL, Phaeo. This file can be read with read_chl.c. P.cdf - has the absorption coefficients of the particulates (ap) and detritus (ad) for 300-750 nm. The method of estimating ap: Mitchell, B.G., Ocean Optics X, p.137-148, 1990; method of estimating ad: Kishino et al. 1985. This file can be read with read_P.c. S.cdf - has the absorption coefficients of CDOM (ag) for 250-650 nm. This file can be read with read_S.c. Some values at longer wavelengths may be small negative numbers: this is the result of small errors when subtracting the blank or doing the red correction. A more robust method of estimating ag at a particular wavelength is not to use the measured value but to fit an exponetial decay curve through the data and using the value from the fitted curve. We do this regularly and can provide the estimated ag value at 300 nm, the exponential slope and the error of the fit. The exponential slope seems to change for wavelengths below 300 nm; therefore we do not use values <300 nm for fitting the exponential curve in the visible range.