Total alkalinity of seawater (TA): the total net concentration of hydrogen ion acceptors in seawater. Since the pH of natural seawater is around 8, the weak acid anions that affect total alkalinity are mainly HCO3-, CO32- and B(OH)4- , so TA≈[HCO3-]+2[CO32-]+[B(OH)4-]+[OH-]-[H+]. The unit is μmol/kg.
Seawater pH (pH): an important indicator to characterize the degree of acidification of seawater. The pH standard used in this product is total scale (pHT=-log([H+]F+[HSO4-])=-log[H+]T). This is a dimensionless parameter.
Seawater dissolved inorganic carbon concentration (DIC): refers to the concentration of inorganic carbon dissolved in water. It is defined as DIC=[CO2]+[HCO3-]+[CO32-], with the unit of μmol/kg.
Seawater CO2 partial pressure (pCO2): Under a certain temperature and salinity, the carbon dioxide partial pressure of seawater and air reaches equilibrium. The unit is μatm.
This algorithm starts from the calculation of the ocean carbonate system. It first solves the problem of insufficient global measured data and constructs a massive navigation observation/calculation data set with four parameter matching. On this basis, through a large number of training and testing experiments, the machine learning algorithm (XGBOOST algorithm) suitable for big data mining was selected and confirmed, as well as the data parameter combination strategy, namely three-band remote sensing reflectance (Rrs413, 443, 488nm), chlorophyll concentration (Chl), and sea surface temperature (SST), sea surface salinity (SSS), mixed layer depth (MLD), sea surface atmospheric pressure (SLP), atmospheric CO2 concentration (xCO2) and same-latitude temperature difference (SST_lat), a total of 9 input parameters. The selection of input parameters takes into account the ecological environment and water mass changes in different sea areas around the world. Mixed layer depth (MLD) and same-latitude temperature difference (SST_lat) are used to characterize the impact of seasonal mixing and upwelling on carbonate parameters. xCO2 is used to characterize the global atmosphere. The impact of continued CO2 growth on seawater pCO2; modeling of open oceans and marginal seas respectively. Finally, pCO2, TA, DIC and pH remote sensing models suitable for the global ocean were constructed, and a 4km monthly average long-term series gridded product from 2003 to 2019 was produced.
unfold
