TL;DR
This paper investigates the drivers of year-to-year fluctuations in the global land carbon sink, finding that while water availability is the dominant global control, different modeling approaches disagree on the role of temperature, particularly in the Northern Hemisphere. This disagreement stems from how well models capture seasonal temperature compensation effects, which is crucial for understanding future carbon-climate feedbacks.
What Problem Does This Paper Address?
The Earth's land surface acts as a significant carbon sink, absorbing a substantial portion of anthropogenic CO2 emissions and thus slowing global warming. However, the strength of this sink fluctuates annually, leading to variations in the rate of atmospheric CO2 increase. Understanding the drivers of this interannual variability (IAV) in the net ecosystem exchange (NEE<sub>IAV</sub>) is critical for predicting future atmospheric CO2 levels and the strength of carbon-climate feedbacks. Previous research has presented conflicting views on whether terrestrial water storage (TWS) or temperature (T) is the primary driver of global NEE<sub>IAV</sub>. This debate arises from differences in how various studies assess ecosystem responses across different regions and seasons, leading to a lack of consensus on how these regional and seasonal controls aggregate to influence the global signal.
Methods
The authors employed three distinct approaches to quantify the regional and seasonal contributions to global NEE<sub>IAV</sub>:
- Atmospheric Inversions: These models use atmospheric CO2 measurements to infer land-atmosphere carbon fluxes. The study utilized data from CAMS and Jena CarboScope models, including a specialized NEE-T-TWS inversion designed to disentangle temperature and water storage effects.
- Process-Based Vegetation Models (DGVMs): These models simulate ecosystem processes based on physical and biological principles. The study used data from the TRENDY project's version 7, which includes 14 different DGVMs.
- Data-Driven Models (FLUXCOM): These models use machine learning to upscale eddy-covariance flux measurements from ground sites to a global scale, driven by satellite and meteorological data.
To analyze the drivers, the researchers calculated correlations between annual global NEE<sub>IAV</sub> and TWS/T. They then used partial correlations to isolate the independent effects of T and TWS. Furthermore, they disaggregated these contributions by region (tropics vs. Northern Hemisphere) and season (boreal spring, summer, autumn, winter) to understand how local and seasonal controls aggregate to the global signal.
Datasets
The study utilized several key datasets:
- Net Ecosystem Exchange (NEE):
- Atmospheric inversion products (CAMS, Jena CarboScope, NEE-T-TWS) for 1979-2016.
- Dynamic Global Vegetation Model (DGVM) outputs from the TRENDY project (version 7) for 1979-2016.
- FLUXCOM global carbon flux dataset (RS+METEO) for 1979-2016.
- Terrestrial Water Storage (TWS): Reconstructed TWS data based on GRACE satellite observations for 1979-2016. Simulated soil moisture from DGVMs and water availability index (WAI) from FLUXCOM were also used for comparison.
- Temperature (T): Gridded monthly air temperature data from the Climatic Research Unit (CRU TS4.02) for 1979-2016.
- CO2 Growth Rate (CGR): Atmospheric CO2 mole fraction data from NOAA/ESRL for 1979-2016, used as a proxy for global NEE<sub>IAV</sub>.
Key Insights and Ideas
- Tropical Dominance: All three modeling approaches agree that tropical ecosystems contribute significantly (over 63%) to the global correlations between NEE<sub>IAV</sub> and TWS/T. This highlights the outsized role of tropical regions in global carbon cycle variability.
- The TWS vs. Temperature Debate: While inversions and DGVMs suggest TWS is the dominant driver of global NEE<sub>IAV</sub>, data-driven FLUXCOM models point to temperature. This divergence is a central finding and a key area of uncertainty.
- Northern Hemisphere Seasonal Compensation is Key: The disagreement between models largely stems from how they represent seasonal temperature effects in the Northern Hemisphere (NH). Inversions and DGVMs show inter-seasonal compensation (e.g., warm spring uptake offset by reduced summer uptake), leading to a TWS dominance. FLUXCOM models exhibit weaker compensation, resulting in a temperature dominance.
- Tropical Seasonality: While tropics are crucial, the study found that contributions from both dry and wet seasons are roughly equal, suggesting that understanding tropical carbon dynamics requires considering both periods.
- Data-Driven Model Limitations: The FLUXCOM model's temperature dominance is linked to its weaker representation of seasonal compensation in the NH, which is not well-supported by local flux tower data. This suggests potential issues with how FLUXCOM extrapolates temperature sensitivities.
Key Results
- Global Drivers: At the annual scale, inversions and DGVMs show a stronger correlation between global NEE<sub>IAV</sub> and TWS than with T. FLUXCOM models, however, show the opposite, with a stronger correlation to T.
- Regional Contributions: Tropics contribute over 63% to global correlations. Northern Hemisphere contributions differ significantly between model types, particularly in the magnitude of temperature's influence.
- Northern Hemisphere Temperature Effects: Inversions and DGVMs show a negative temperature sensitivity in boreal spring (warmer spring, more uptake) and a positive sensitivity in summer (warmer summer, less uptake), leading to compensation. FLUXCOM models show a positive spring sensitivity and a stronger positive summer sensitivity, resulting in less compensation.
- Tropical Contributions: Tropical Africa and America generally contribute more to global correlations than tropical Asia, though uncertainties remain.
Limitations and Future Work
- Model Disagreement: The primary limitation is the persistent disagreement between the three modeling approaches regarding the dominant driver of global NEE<sub>IAV</sub>, particularly concerning the role of temperature in the NH. This highlights the need for further refinement of all model types.
- NH Seasonal Compensation: The study emphasizes that the accuracy of NH seasonal compensation effects is crucial. Future work should focus on improving process-based models and data-driven approaches to better capture these dynamics, especially in light of potential future changes due to climate change (e.g., reduced chilling, permafrost thaw).
- Tropical NEE Estimation: The accuracy of tropical NEE estimates from atmospheric inversions and FLUXCOM models is limited by sparse ground-based observations. Future efforts should focus on assimilating satellite CO2 measurements and improving DGVMs' representation of key processes like nutrient limitations and soil hydrology.
- Data-Driven Model Training: The FLUXCOM model's reliance on training data from temperate regions might influence its extrapolation to high-latitude ecosystems. Further investigation into how training data influences predictions is warranted.
- Roadmap for Future Research: The paper provides a roadmap for future measurement campaigns and model development to reduce uncertainties in understanding large-scale NEE variations and their dominant drivers.
Summary updated