Experiment

EXP 04: Climatological Testbed

Tests whether the early-2000s ocean carbon sink hiatus can be explained by sampling alone.

Overview

Many flux data products show a hiatus in the ocean carbon sink in the early 2000s. This experiment tests whether that signal could be an artefact of sparse sfCO₂ observations. It uses a synthetic sfCO₂ field without interannual variability, so any interannual variability in the sfCO₂ reconstructions can only come from the sampling.

Lead Author

Manuscript

In preparation

Experiment Information

Status:In preparation
DOI:To be added
Testbed: Synthetic sfCO₂ field based on the Takahashi climatology (Takahashi et al., 2009):
  • No-trend: identical climatology every year
  • With-trend: trend of 2 µatm yr⁻¹ added to sfCO₂ and atmospheric xCO₂
sfCO₂ reconstruction methods: 8 sfCO₂ reconstruction methods:
  • CarboScope (Rödenbeck et al., 2014, 2022)
  • CMEMS-LSCE-v1 (Chau et al., 2022)
  • JMA-MLR (Iida et al., 2021)
  • LDEO pCO2-Residual (Bennington et al., 2022)
  • NIES-ML3 (Zeng et al., 2022)
  • OceanSODA-ETHZv2 (Gregor et al., 2024)
  • UExP-FNN-U (Watson et al., 2020; Ford et al., 2024)
  • VLIZ-SOMFFN (Landschützer et al., 2016)
SOCAT mask used:SOCATv2023
Design:EXP 04 evaluates the skill of sfCO₂ reconstruction methods in a simple testbed and assesses the impact of sparse sfCO₂ observations. Specifically, the motivation is to answer the question: is the early-2000s hiatus in the ocean carbon sink, shown by many flux data products, attributable to sampling alone, or is it a true shift in carbon uptake?

In this experiment, the true sfCO₂ field has no interannual variability and no internal physical or biogeochemical dynamics. The sfCO₂ reconstructions therefore only aim to identify the prescribed seasonal cycle and long-term mean sfCO₂, given the available sampling density. Two sub-experiments are conducted: a ‘no-trend’ experiment, in which the climatology is identical each year, and, to more closely replicate real-world conditions, a ‘with-trend’ experiment, which adds a trend of 2 µatm yr⁻¹ to both the Takahashi sfCO₂ climatology and the atmospheric xCO₂ field. All other predictor variables contain no trend.