Feasibility of reconstructing the summer basin-scale sea surface partial pressure of carbon dioxide from sparse in situ observations over the South China Sea

Abstract

Sea surface partial pressure of CO2 (pCO2) data with a high spatiotemporal resolution are important in studying the global carbon cycle and assessing the oceanic carbon uptake. However, the observed sea surface pCO2 data are usually limited in spatial and temporal coverage, especially in marginal seas. This study provides an approach to reconstruct the complete sea surface pCO2 field in the South China Sea (SCS) with a grid resolution of 0.5∘×0.5∘ over the period of 2000–2017 using both remote-sensing-derived pCO2 and observed underway pCO2, among which the gridded underway pCO2 data in 2004, 2005, and 2006 are presented for the first time. Empirical orthogonal functions (EOFs) were computed from the remote-sensing-derived pCO2. Then, a multilinear regression was applied to the observed pCO2 as the response variable with the EOFs as the explanatory variables. EOF1 explains the general spatial pattern of pCO2 in the SCS. EOF2 shows the pattern influenced by the Pearl River plume on the northern shelf and slope. EOF3 is consistent with the pattern influenced by coastal upwelling along the northern coast of the SCS. When pCO2 observations cover a sufficiently large area, the reconstructed fields successfully display a pattern of relatively high pCO2 in the mid and southern basin. The rate of sea surface pCO2 increase in the SCS is 2.4±0.8 µatm yr−1 based on the spatial average of the reconstructed pCO2 over the period of 2000–2017. This is consistent with the temporal trends at Station SEATS (SouthEast Asia Time-series Study; 18∘ N, 116∘ E) in the northern basin of the SCS and at Station ALOHA (A Long-Term Oligotrophic Habitat Assessment; 22∘45′ N, 158∘ W) in the North Pacific. We validated our reconstruction with a leave-one-out cross-validation approach, which yields the root-mean-square error (RMSE) in the range of 2.4–5.2 µatm, smaller than the spatial standard deviation of our reconstructed data and much smaller than the spatial standard deviation of the observed underway data. The RMSE between the reconstructed summer pCO2 and the observed underway pCO2 is no larger than 31.7 µatm, in contrast to (a) the RMSE from 12.8 to 89.0 µatm between the remote-sensing-derived pCO2 and the underway data and (b) the RMSE from 32.6 to 44.5 µatm between the neural-network-produced pCO2 and the underway data. The difference between the reconstructed pCO2 and those calculated from observations at Station SEATS is in the range from −7 to 10 µatm. These comparison results indicate the reliability of our reconstruction method and output. All the data for this paper are openly and freely available at PANGAEA under the link https://doi.org/10.1594/PANGAEA.921210 (Wang et al., 2020).