2.2.2. Remote Sensing Data and Reanalysis Data

Co-temporal and co-located met-ocean parameters that are considered to be somehow connected with the ocean carbonate system were derived from independent sources using earth observation satellite remote sensing (BD 1–7; PD 1–5), model reanalyses (PD 10), and empirical algorithms (PD 6–8) (Table 2). The GEBCO bathymetry (PD 9) was derived

from a mix of ship track soundings, with the interpolation between soundings guided by satellite-derived gravity data.

**Table 2.** Co-temporal and co-located environmental drivers derived from independent sources that range from satellite remote sensing and model analyses to empirical algorithms were collected.


2.2.3. Justification for the Selection and Use of Environmental Drivers

Figure 2 shows the linkage between the various environmental drivers used in this study and how these were used to model the target ocean surface DIC, TA, and pH. The environmental drivers can be seen to represent the following three proxies of oceanic processes:


**Figure 2.** Linkage between the various groups of environmental drivers and how these were used to model or predict the three target parameters of surface DIC, TA, and pH. The environmental drivers can be seen as representing some of the main met-ocean processes influencing these three target variables (based on [38]).

These four processes were used to closely represent as much as possible the forcing that leads to the derivation of DIC, TA, and pH using our algorithm. Native resolution grids of all of the environmental drivers considered for this study, including PD1, were resampled to a common 0.04◦ × 0.04◦ global raster grid for a suitable retrieval of all colocated data. Table 3 provides a summarized justification for the inclusion of these drivers into the predictive algorithm.

**Table 3.** Justification of the use of the biological and physical drivers of surface DIC, TA, and pH used for this study.



#### **Table 3.** *Cont.*

#### *2.3. Algorithm Development and Validation*
