A remote sensing algorithm for planktonic dimethylsulfoniopropionate (DMSP) and an analysis of global patterns

作者：Gali Marti; Devred Emmanuel; Levasseur Maurice; Royer Sarah Jeanne; Sabin Marcel

来源：Remote Sensing of Environment, 2015, 171: 171-184.

DOI：10.1016/j.rse.2015.10.012

• Dimethylsulfoniopropionate (DMSP)
• Sulfur cycle
• Phytoplankton
• Chlorophyll
• Stratification
• Particulate inorganic carbon (PIC)
• Algorithm

Dimethylsulfoniopropionate (DMSP) is a ubiquitous phytoplankton metabolite and the main precursor of the climate-active gas dimethylsulfide (DMS) in the oceans' surface. Here we use total DMSP (DMSPt) and ancillary measurements from a global database to develop a remote sensing algorithm for DMSPt in the upper mixed layer (UML). Over 55% of total DMSPt variability (log(10) scale) is explained by in situ chlorophyll a (Chl) after dividing the database into two subsets, according to "stratified" and "mixed" water column criteria, based on the ratio between euphotic layer depth (Z(eu)) and mixed layer depth (MLD). Up to 70% of the variability is explained when adding sea surface temperature (SST) and log(10)(Z(eu)/MLD) as predictors for the stratified and mixed subsets, respectively. Independent validation on satellite Chl match-ups indicates that the algorithm predicts DMSPt across three orders of magnitude with a root-mean-squared error spanning from 0.20 to 0.26 (log(10) space) and mean absolute error typically around 45% (linear space). An additional submodel based on remotely sensed particulate inorganic carbon (PIC) is used to predict DMSPt in coccolithophore blooms if satellite Chl is not available. We use the algorithm to produce a monthly global DMSPt climatology, and estimate that DMSP synthesis amounts to 5-9% of oceanic phytoplankton gross carbon production. Our algorithm provides a new remote sensing tool for resolving temporal and spatial variations in DMSPt concentration, and represents a step forward toward improved diagnosis of contemporary DMS emission based on satellite Earth observation.

• 出版日期2015-12-15