Estimating temporal and spatial variation of ocean surface pCO2 in the North Pacific using a self-organizing map neural network technique

[摘要] This study uses a neural network technique to produce maps of the partialpressure of oceanic carbon dioxide (pCO₂^sea) in the NorthPacific on a 0.25° latitude × 0.25° longitude gridfrom 2002 to 2008. The pCO₂^sea distribution was computed usinga self-organizing map (SOM) originally utilized to map thepCO₂^sea in the North Atlantic. Four proxy parameters – seasurface temperature (SST), mixed layer depth, chlorophyll a concentration,and sea surface salinity (SSS) – are used during the training phase to enablethe network to resolve the nonlinear relationships between thepCO₂^sea distribution and biogeochemistry of the basin. Theobserved pCO₂^sea data were obtained from an extensive datasetgenerated by the volunteer observation ship program operated by the NationalInstitute for Environmental Studies (NIES). The reconstructedpCO₂^sea values agreed well with the pCO₂^seameasurements, with the root-mean-square error ranging from 17.6 μatm(for the NIES dataset used in the SOM) to 20.2 μatm (forindependent dataset). We confirmed that the pCO₂^sea estimatescould be improved by including SSS as one of the training parameters and bytaking into account secular increases of pCO₂^sea that havetracked increases in atmospheric CO₂. Estimated pCO₂^seavalues accurately reproduced pCO₂^sea data at several timeseries locations in the North Pacific. The distributions ofpCO₂^sea revealed by 7 yr averaged monthlypCO₂^sea maps were similar to Lamont-Doherty Earth ObservatorypCO₂^sea climatology, allowing, however, for a more detailedanalysis of biogeochemical conditions. The distributions ofpCO₂^sea anomalies over the North Pacific during the winterclearly showed regional contrasts between El Niño and La Niña yearsrelated to changes of SST and vertical mixing.

[学科分类] 地球化学与岩石