Portal de Periódicos da CAPES

The remote sensing of ocean primary productivity: Use of a new data compilation to test satellite algorithms

1992; American Geophysical Union; Volume: 97; Issue: C2 Linguagem: Inglês

10.1029/91jc02843

2156-2202

William M. Balch, Robert H. Evans, Jim Brown, Gene C. Feldman, Charles R. McClain, Wayne E. Esaias,

Isotope Analysis in Ecology

We tested global pigment and primary productivity algorithms based on a new data compilation of over 12,000 stations occupied mostly in the northern hemisphere, from the late 1950's to 1988. The results showed high variability of the fraction of total pigment contributed by chlorophyll a (ρ), which is required for subsequent predictions of primary productivity. Two models, which predict pigment concentration normalized to attenuation length or euphotic depth, were checked against 2,800 vertical profiles of pigments (chlorophyll a , phaeopigment and total pigment). Phaeopigments consistently showed maxima at about one optical depth below the chlorophyll maxima. We also checked the global Coastal Zone Color Scanner (CZCS; daily 20km resolution) archive for data coincident with the sea truth data. A regression of satellite‐derived pigment versus ship‐derived pigment had a coefficient of determination ( r 2 ) of 0.40 (n=731 stations). The satellite underestimated the true pigment concentration in mesotrophic and oligotrophic waters (< 1 mg pigment m −3 ) and overestimated the pigment concentration in eutrophic waters (> 1 mg pigment m‐3). The error in the satellite estimate showed no trends with time between 1978 and 1985. In general the variability of the satellite retrievals increased with pigment concentration. Several productivity algorithms were tested which utilize information on the photoadaptive parameters, biomass and optical parameters for predicting integral production. The most reliable algorithm which explained 67% of the variance in integral production for 1676 stations suggested that future success in deriving primary productivity from remotely sensed data will rely on accurate retrievals of “living” biomass from satellite data, as well as the prediction of at least one photoadaptive parameter such as maximum photosynthesis.