Trends and drivers of regional sources and sinks of carbon dioxide over the past two decades

Stephen Sitch, Pierre Friedlingstein, Nicolas Gruber, S. D. M. Jones, Guillermo N. Murray‐Tortarolo, Anders Ahlström, Scott C. Doney, Heather Graven, Christoph Heinze, Chris Huntingford, Samuel Levis, Peter Levy, M. Lomas, Benjamin Poulter, Nicolas Viovy, Sönke Zaehle, Ning Zeng, Almut Arneth, Gordon B. Bonan, Laurent Bopp, Josep G. Canadell, Frédéric Chevallier, Philippe Ciais, Rich Ellis, Manuel Gloor, Philippe Peylin, Shilong Piao, Corinne Le Quéré, Benjamin Smith, Zhiliang Zhu, Ranga B. Myneni2013 · DOI

TL;DR

This study quantifies and attributes trends in regional carbon dioxide (CO2) sinks on land and in the ocean from 1990-2009 using a suite of sophisticated climate and vegetation models. It reveals that while the land sink is increasing, primarily due to CO2 fertilization of plant growth in the tropics, the ocean sink shows a much weaker trend, counteracted by warming sea surface temperatures.

What Problem Does This Paper Address?

The Earth's land and oceans act as crucial "sinks," absorbing over half of the CO2 emitted by human activities each year. Understanding how these sinks are changing and what drives these changes is vital for predicting future atmospheric CO2 concentrations and climate trajectories. While global carbon budgets are becoming more constrained, significant uncertainties remain regarding the regional distribution and trends of these sinks, and the underlying processes (like climate variability and CO2 fertilization) that modulate them. This paper aims to fill this gap by providing a comprehensive, multi-model assessment of regional carbon exchange trends and their drivers over the past two decades.

Methods

The researchers employed a "bottom-up" modeling approach, utilizing:

  • Nine Dynamic Global Vegetation Models (DGVMs): These models simulate the carbon and hydrological cycles of terrestrial ecosystems, representing processes like photosynthesis, respiration, and carbon storage.
  • Four Ocean Biogeochemical General Circulation Models (OBGCMs): These models simulate the exchange of CO2 between the atmosphere and the ocean, considering factors like ocean circulation, temperature, and biological activity.

The models were driven by reconstructed climate fields and observed atmospheric CO2 concentrations. Crucially, the study performed "factorial simulations" where specific drivers (like CO2 concentration or climate) were varied independently. This allowed the researchers to attribute observed trends to underlying processes, such as CO2 fertilization, climate effects on productivity and respiration, and ocean warming. Land Use and Land Cover Change (LULCC) was not explicitly included in the DGVM simulations for this study, but its potential influence is discussed.

Datasets

  • Reconstructed climate fields: Used to force the DGVMs.
  • Observed global atmospheric CO2 concentrations: Used as a forcing for both land and ocean models.
  • Remote sensing data (e.g., Leaf Area Index - LAI): Used to evaluate trends in land greenness.
  • Atmospheric inversion results: Used for comparison with modeled land sink trends.
  • Oceanic pCO2, DIC, alkalinity, temperature, salinity, and mixed layer depth data: Used for ocean model evaluation.

Key Insights and Ideas

  • CO2 Fertilization is a Major Land Sink Driver: The increasing land sink is largely driven by enhanced Net Primary Production (NPP), which is strongly influenced by rising atmospheric CO2 levels. This highlights the significant, though not unlimited, role of CO2 fertilization in boosting plant growth.
  • Tropical Land is Key to the Increasing Land Sink: The tropics are identified as the primary source of the trend towards an increasing land sink, suggesting these regions are playing a disproportionately large role in absorbing atmospheric CO2.
  • Ocean Sink Trends are Weak and Complex: Unlike the land, the ocean sink shows a much smaller trend over the study period. This is because the positive effect of increasing atmospheric CO2 on ocean uptake is largely counteracted by climate variability and change, particularly warming sea surface temperatures.
  • Regional Trends Vary Significantly: While global trends are discernible, regional patterns are more complex. Some land regions show decreasing sink trends, potentially indicating a future shift towards becoming carbon sources if drying trends continue.
  • Model Agreement is Encouraging but Uncertainties Remain: The study highlights areas of good agreement among models, particularly for global trends. However, significant uncertainties persist at regional scales, especially concerning the influence of land use change and the representation of complex ecosystem processes.

Key Results

  • Global Land Sink: Simulated mean sink of -2.4 ± 0.7 Pg C yr⁻¹ with a significant increasing trend of -0.06 ± 0.03 Pg C yr⁻².
  • Global Ocean Sink: Simulated mean sink of -2.2 ± 0.2 Pg C yr⁻¹ (1990-2004) with a trend indistinguishable from zero (-0.01 ± 0.02 Pg C yr⁻²). Models extending to 2009 showed a slightly stronger trend of -0.02 ± 0.01 Pg C yr⁻².
  • Land Sink Drivers: Trend in land sink driven by increasing NPP (0.22 ± 0.08 Pg C yr⁻²) exceeding the trend in heterotrophic respiration (0.16 ± 0.05 Pg C yr⁻²).
  • Tropical Land Contribution: The majority of the land-based trend originates from tropical ecosystems (-0.04 ± 0.01 Pg C yr⁻²).
  • Northern Land Trends: Almost no trend in the northern land region, where warming and reduced rainfall offset CO2 fertilization effects.

Limitations and Future Work

  • Land Use and Land Cover Change (LULCC): The DGVMs did not explicitly simulate LULCC, which is acknowledged as a significant factor influencing regional carbon budgets. Future work should incorporate consistent LULCC implementation.
  • Disturbance Representation: Few land models include prognostic wildfire schemes, and none represent other disturbances and their interactions comprehensively. This may lead to conservative estimates of warming and drought impacts.
  • Nitrogen Cycle: Only a few models included interactive nutrient cycling, limiting the ability to fully assess nitrogen's role in constraining CO2 responses.
  • Ocean Model Resolution: Ocean models are often too coarse to adequately represent mesoscale processes like eddies, which can influence ocean carbon dynamics.
  • Ecosystem Process Representation: Ocean ecosystem modules need improvement to better characterize responses to climate change and other stressors like acidification.
  • Data Limitations: Evaluation of advanced model components is hindered by data availability.

The authors emphasize the need for improved model benchmarking, updated datasets, and more comprehensive representation of key processes in both land and ocean models to reduce uncertainties and better predict future carbon cycle changes.

Summary updated

Abstract

The land and ocean absorb on average over half of the anthropogenic emissions of carbon dioxide (CO2) every year. These CO2 "sinks" are modulated by climate change and variability. Here we use a suite of nine Dynamic Global Vegetation Models (DGVMs) and four Ocean Biogeochemical General Circulation Models (OBGCMs) to quantify the global and regional climate and atmospheric CO2 – driven trends in land and oceanic CO2 exchanges with the atmosphere over the period 1990–2009, attribute these trends to underlying processes, and quantify the uncertainty and level of model agreement. The models were forced with reconstructed climate fields and observed global atmospheric CO2; Land Use and Land Cover Changes are not included for the DGVMs. Over the period 1990–2009, the DGVMs simulate a mean global land carbon sink of −2.4 ± 0.7 Pg C yr−1 with a small significant trend of −0.06 ± 0.03 Pg C yr−2 (increasing sink). Over the more limited period 1990–2004, the ocean models simulate a mean ocean sink of –2.2 ± 0.2 Pg C yr–1 with a trend in the net C uptake that is indistinguishable from zero (−0.01 ± 0.02 Pg C yr−2). The two ocean models that extended the simulations until 2009 suggest a slightly stronger, but still small trend of −0.02 ± 0.01 Pg C yr−2. Trends from land and ocean models compare favourably to the land greenness trends from remote sensing, atmospheric inversion results, and the residual land sink required to close the global carbon budget. Trends in the land sink are driven by increasing net primary production (NPP) whose statistically significant trend of 0.22 ± 0.08 Pg C yr−2 exceeds a significant trend in heterotrophic respiration of 0.16 ± 0.05 Pg C yr−2 – primarily as a consequence of wide-spread CO2 fertilisation of plant production. Most of the land-based trend in simulated net carbon uptake originates from natural ecosystems in the tropics (−0.04 ± 0.01 Pg C yr−2), with almost no trend over the northern land region, where recent warming and reduced rainfall offsets the positive impact of elevated atmospheric CO2 on carbon storage. The small uptake trend in the ocean models emerges because climate variability and change, and in particular increasing sea surface temperatures, tend to counteract the trend in ocean uptake driven by the increase in atmospheric CO2. Large uncertainty remains in the magnitude and sign of modelled carbon trends in several regions, and on the influence of land use and land cover changes on regional trends.
BibTeX
@misc{sitch2013trends,
  author = {Stephen Sitch and Pierre Friedlingstein and Nicolas Gruber and S. D. M. Jones and Guillermo N. Murray‐Tortarolo and Anders Ahlström and Scott C. Doney and Heather Graven and Christoph Heinze and Chris Huntingford and Samuel Levis and Peter Levy and M. Lomas and Benjamin Poulter and Nicolas Viovy and Sönke Zaehle and Ning Zeng and Almut Arneth and Gordon B. Bonan and Laurent Bopp and Josep G. Canadell and Frédéric Chevallier and Philippe Ciais and Rich Ellis and Manuel Gloor and Philippe Peylin and Shilong Piao and Corinne Le Quéré and Benjamin Smith and Zhiliang Zhu and Ranga B. Myneni},
  title = {{Trends and drivers of regional sources and sinks of carbon dioxide over the past two decades}},
  year = {2013},
  doi = {10.5194/bgd-10-20113-2013},
}
Figure