A comparative assessment of the uncertainties of global surface ocean CO&amp;lt;sub&amp;gt;2&amp;lt;/sub&amp;gt; estimates using a machine-learning ensemble (CSIR-ML6 version 2019a) – have we hit the wall?

Luke Gregor,Pedro M. Scheel Monteiro,Alice D. Lebehot,Schalk Kok

doi:10.5194/gmd-12-5113-2019

Abstract

Abstract. Over the last decade, advanced statistical inference and machine learning have been used to fill the gaps in sparse surface ocean CO2 measurements (Rödenbeck et al., 2015). The estimates from these methods have been used to constrain seasonal, interannual and decadal variability in sea–air CO2 fluxes and the drivers of these changes (Landschützer et al., 2015, 2016; Gregor et al., 2018). However, it is also becoming clear that these methods are converging towards a common bias and root mean square error (RMSE) boundary: “the wall”, which suggests that pCO2 estimates are now limited by both data gaps and scale-sensitive observations. Here, we analyse this problem by introducing a new gap-filling method, an ensemble average of six machine-learning models (CSIR-ML6 version 2019a, Council for Scientific and Industrial Research – Machine Learning ensemble with Six members), where each model is constructed with a two-step clustering-regression approach. The ensemble average is then statistically compared to well-established methods. The ensemble average, CSIR-ML6, has an RMSE of 17.16 µatm and bias of 0.89 µatm when compared to a test dataset kept separate from training procedures. However, when validating our estimates with independent datasets, we find that our method improves only incrementally on other gap-filling methods. We investigate the differences between the methods to understand the extent of the limitations of gap-filling estimates of pCO2. We show that disagreement between methods in the South Atlantic, southeastern Pacific and parts of the Southern Ocean is too large to interpret the interannual variability with confidence. We conclude that improvements in surface ocean pCO2 estimates will likely be incremental with the optimisation of gap-filling methods by (1) the inclusion of additional clustering and regression variables (e.g. eddy kinetic energy), (2) increasing the sampling resolution and (3) successfully incorporating pCO2 estimates from alternate platforms (e.g. floats, gliders) into existing machine-learning approaches.

Highlights

The ocean plays a crucial role in mitigating against climate change by taking up about a third of anthropogenic carbon dioxide (CO2) emissions (Sabine et al, 2004; Khatiwala et al, 2013; McKinley et al, 2016)
The results from the regression comparisons are depicted in Fig. 5a–c, which plot the matrix of the (a) average bias, (b) root mean square error (RMSE) and (c) relative interannual variability metric (Riav) for each combination of the experimental number of clusters and clustering features
Our study suggests that we may be reaching the limits of gapfilling methods’ abilities to reduce uncertainties, as shown by the limited incremental improvement in errors by the ensemble method we compare with established methods

Summary

Introduction

The ocean plays a crucial role in mitigating against climate change by taking up about a third of anthropogenic carbon dioxide (CO2) emissions (Sabine et al, 2004; Khatiwala et al, 2013; McKinley et al, 2016). While the mean state in the global contemporary marine CO2 uptake is a widely used benchmark (Le Quéré et al, 2018), underlying assumptions and limited confidence regarding the variability and longterm evolution of this sink persist. Sparse observations of surface ocean CO2 during winter and in large inaccessible regions have been the biggest barrier in constraining the seasonal and interannual variability of global contemporary sea–air exchange (Monteiro et al, 2010; Rödenbeck et al, 2015; Bakker et al, 2016; Ritter et al, 2017). The increasing ship-based sampling effort and the ongoing development of autonomous observational platforms (e.g. biogeochemical Argo floats and Wavegliders) have improved confidence of interannual estimates of ocean CO2 uptake in more recent. L. Gregor et al.: CSIR-ML6 version 2019a years (Monteiro et al, 2015; Bakker et al, 2016; Gray et al, 2018)

Methods

Discussion

Conclusion