Optimum satellite remote sensing of the marine carbonate system using empirical algorithms in the global ocean, the Greater Caribbean, the Amazon Plume and the Bay of Bengal

Abstract

Improving our ability to monitor ocean carbonate chemistry has become a priority as the ocean continues to absorb carbon dioxide from the atmosphere. This long-term uptake is reducing the ocean pH; a process commonly known as ocean acidification. The use of satellite Earth Observation has not yet been thoroughly explored as an option for routinely observing surface ocean carbonate chemistry, although its potential has been highlighted. We demonstrate the suitability of using empirical algorithms to calculate total alkalinity (A T) and total dissolved inorganic carbon (C T), assessing the relative performance of satellite, interpolated in situ, and climatology datasets in reproducing the wider spatial patterns of these two variables. Both A T and C T in situ data are reproducible, both regionally and globally, using salinity and temperature datasets, with satellite observed salinity from Aquarius and SMOS providing performance comparable to other datasets for the majority of case studies. Global root mean squared difference (RMSD) between in situ validation data and satellite estimates is 17 μmol kg −1 with bias < 5 μmol kg −1 for A T and 30 μmol kg −1 with bias < 10 μmol kg −1 for C T. This analysis demonstrates that satellite sensors provide a credible solution for monitoring surface synoptic scale A T and C T. It also enables the first demonstration of observation-based synoptic scale A T and C T temporal mixing in the Amazon plume for 2010-2016, complete with a robust estimation of their uncertainty.

OceanSODA-ETHZ: a global gridded data set of the surface ocean carbonate system for seasonal to decadal studies of ocean acidification

“…At the global scale, most of what we know about the progression of ocean acidification in the recent decades has come from either models (Bopp et al, 2013;Kwiatkowski et al, 2020) or from the combination of model-based trends with observation-based climatologies (Feely et al, 2009;Jiang et al, 2019). A notable exception is the large number of studies that have analyzed the trends and variability of surface ocean pCO 2 (e.g., Landschützer et al, 2013Landschützer et al, , 2016Rödenbeck et al, 2014;Denvil-Sommer et al, 2019;Gregor et al, 2019) and the effort of Lauvset et al (2015) and Turk et al (2017) to analyze long-term trends in pH and respectively. But these studies remained limited to one single parameter.…”

“…By far the most established efforts are those that interpolate and extrapolate the ocean pCO 2 observations, as demonstrated by the intercomparison project by Rödenbeck et al (2015). Feed-forward neural networks (FFNNs) have become one of the favored tools (Landschützer et al, 2013;Zeng et al, 2014;Denvil-Sommer et al, 2019), but other statistical and machine learning methods, such as Bayesian regression and tree-based regression, have also been used with similar success (Rödenbeck et al, 2014;Gregor et al, 2019). However, the specific implementation of the methods is what sets the assortment of methods apart.…”

“…However, the specific implementation of the methods is what sets the assortment of methods apart. For example, the SOM-FFN method of Landschützer et al (2013) and the CSIR-ML6 method of Gregor et al (2019) (among others) first cluster the data based on a certain set of climatological predictors and then perform a regression on pCO 2 for each resulting cluster. An alternate approach, used by both the CMEMS-FFNNv2 (Denvil-Sommer et al, 2019) and NIES-FNN (Zeng et al, 2014) methods, is to include the positional coordinates, without the need for subsetting the data by clustering.…”

OceanSODA-ETHZ: A global gridded data set of the surface ocean carbonate system for seasonal to decadal studies of ocean acidification

“…inter-and extrapolation effort. Advances in remote sensing (Land et al, 2019), and the increasing power and usability of machine learning techniques have permitted to address this challenge, leading to a proliferation of such efforts. But they vary greatly between the different parameters of the marine carbonate system.…”

“…A similar RMSE was achieved by Broullón et al (2018) using a neural network approach (NNGv2). In addition to these global regressions, several regionally specific regressions were developed (see Table 1 in Land et al 2019).…”