Evaluating the Interplay Between Biophysical Processes and Leaf Area Changes in Land Surface Models

Giovanni Forzieri, Grégory Duveiller, Goran Georgievski, Wei Li, Eddy Robertson, Markus Kautz, Peter Lawrence, Lorea Garcia San Martin, Peter Anthoni, Philippe Ciais, Julia Pongratz, Stephen Sitch, Andy Wiltshire, Almut Arneth, Alessandro CescattiJournal of Advances in Modeling Earth Systems 2018 · DOI

TL;DR

This paper evaluates how well different Land Surface Models (LSMs) capture the relationship between vegetation changes (leaf area) and energy exchange with the atmosphere. They found that models generally perform better for grasses than for trees, especially in colder climates, highlighting key areas for improving how LSMs represent vegetation's role in climate.

What Problem Does This Paper Address?

Land Surface Models (LSMs) are crucial tools for understanding and predicting how Earth's land surface interacts with the climate system. A key aspect of these interactions is how changes in vegetation, particularly leaf area, influence the exchange of energy (like heat and moisture) between the land and the atmosphere. While LSMs are widely used, their ability to accurately represent these complex vegetation-biophysics feedbacks, especially across diverse climates, is not fully understood. This paper aims to fill this knowledge gap by systematically evaluating an ensemble of leading LSMs against satellite observations to pinpoint their strengths and weaknesses in simulating these crucial interactions. This is particularly important given observed global greening trends that could significantly impact regional climates.

Methods

The researchers developed a novel evaluation framework to assess LSM performance. They focused on the interannual variations (year-to-year changes) of Leaf Area Index (LAI) and the components of the surface energy balance (net radiation, latent heat, and sensible/ground heat fluxes). By analyzing the relationships between these variations across different climate gradients (temperature and precipitation), they could identify how well models capture these dynamic feedbacks. They used a "bioclimatic space" approach, binning data based on LAI and climate variables to visualize these relationships. Statistical metrics like percent bias (PBIAS), Root Mean Square Error (RMSE), and Spearman rank correlation were used to quantify model performance.

Datasets

  • Observation-based products:
    • Net radiation (RN): CERES SYN1deg
    • Latent heat (LE): Derived from GLEAM v2b (Global Land Evaporation Amsterdam Model)
    • Sensible and ground heat fluxes (H+G): Calculated from RN and LE.
  • Leaf Area Index (LAI): GIMMS3g (Global Inventory Modeling and Mapping Studies Normalized Difference Vegetation Index)
    • Precipitation (P) and Air Temperature (T): CRU-NCEP product.
  • Plant Functional Types (PFTs): ESA-CCI (European Space Agency's Climate Change Initiative) land cover map.
  • Irrigated Area Fraction: GMIA (Global Map of Irrigation Areas).
  • Land Surface Model (LSM) simulations:
    • JSBACH
    • JULES
    • ORCHIDEE
    • CLM (Community Land Model)
    • LPJ-GUESS

The study used a consistent forcing dataset (CRU-NCEP climate and CO2 concentrations) for all LSMs and ran simulations for pure PFTs to isolate vegetation-specific responses.

Key Insights and Ideas

  • Grasses vs. Trees: Models generally do a better job of capturing the interplay between vegetation and energy fluxes for grasses compared to trees. This suggests that the processes governing grass growth and its impact on energy exchange are better represented.
  • Cold Climate Challenges: A significant and common bias across multiple models is the poor representation of the relationship between LAI and net radiation in cold and temperate climates for trees. This points to issues in how models handle LAI-driven albedo effects in these regions.
  • Transpiration vs. Albedo Dominance: For trees, models often show a dominance of transpiration-driven mechanisms over albedo-driven effects, which contradicts observational evidence. This suggests an imbalance in how these two key processes are represented.
  • LAI Variability is Key: The study highlights that year-to-year variability in LAI is crucial for accurately simulating energy fluxes. Models that underestimate LAI variability struggle to capture the observed relationships.
  • No Single "Best" Model: The evaluation reveals that no single LSM consistently outperforms others across all metrics, vegetation types, and climate conditions. Different models have distinct strengths and weaknesses.

Key Results

  • Grasses: The interplay between turbulent fluxes (latent and sensible heat) and LAI is generally well represented in models for grasses, increasing confidence in their simulation of LAI-dependent partitioning of net radiation.
  • Trees: Models exhibit systematic biases in the relationship between LAI variability and net radiation in cold and temperate climates. This affects the amount of absorbed radiation due to LAI's impact on surface albedo. Furthermore, the relationships between LAI and turbulent fluxes in tree biomes often contradict experimental evidence.
  • Model Performance Metrics:
    • JSBACH: Underestimates LAI variability, particularly in boreal/temperate zones. Struggles with the LAI-net radiation interplay.
  • JULES: Substantially underestimates LAI variability due to its LAI being tied to total vegetation carbon. Shows a negative spatial correlation between latent and sensible heat variability, unlike observations.
  • CLM: Overestimates LAI variability, especially in cold/warm regions under moderately wet conditions. Shows reverse patterns in net radiation vs. LAI in high-latitude boreal areas.
  • ORCHIDEE: Captures LAI variability reasonably well but underestimates it in wet regimes. Shows a strong negative spatial correlation between latent and sensible heat variability, not found in observations.
  • LPJ-GUESS: Captures LAI variability well but overestimates it in warm, moderately wet regimes. Shows opposite changes in latent heat compared to observations in many conditions.
  • Spatial Correlation: Temporal correlation maps reveal that while models might show reasonable agreement at the bioclimatic space level, specific spatial patterns of these relationships can differ significantly from observations.

Limitations and Future Work

  • Model-Specific Limitations: The paper details specific model limitations in Table 3, such as constraints on maximum LAI, limitations in phenology models, and potential issues with snow cover representation.
  • Common Biases: A significant common bias is the misrepresentation of the LAI-net radiation interplay in cold/temperate climates for trees, likely due to imprecise modeling of LAI-related albedo effects. The dominance of transpiration over albedo mechanisms in tree biomes is another recurring issue.
  • Observational Uncertainty: The authors acknowledge that observational data itself has uncertainties, particularly in cold-temperate climates for certain fluxes and net radiation for grasses. This can lead to a degree of uncertainty in assessing model performance.
  • Future Work: The study suggests that further investigation is needed to improve the understanding and modeling capabilities of LSMs. This includes refining the representation of vegetation-biophysics processes, improving the balance between transpiration and albedo mechanisms, and potentially integrating more detailed observational data and techniques to account for instrumental noise. The authors emphasize that identifying the precise causes of model limitations is complex due to the interconnectedness of processes within LSMs.

Summary updated

Abstract

Land Surface Models (LSMs) are essential to reproduce biophysical processes modulated by vegetation and to predict the future evolution of the land-climate system. To assess the performance of an ensemble of LSMs (JSBACH, JULES, ORCHIDEE, CLM, and LPJ-GUESS) a consistent set of land surface energy fluxes and leaf area index (LAI) has been generated. Relationships of interannual variations of modeled surface fluxes and LAI changes have been analyzed at global scale across climatological gradients and compared with those obtained from satellite-based products. Model-specific strengths and deficiencies were diagnosed for tree and grass biomes. Results show that the responses of grasses are generally well represented in models with respect to the observed interplay between turbulent fluxes and LAI, increasing the confidence on how the LAI-dependent partition of net radiation into latent and sensible heat are simulated. On the contrary, modeled forest responses are characterized by systematic bias in the relation between the year-to-year variability in LAI and net radiation in cold and temperate climates, ultimately affecting the amount of absorbed radiation due to LAI-related effects on surface albedo. In addition, for tree biomes, the relationships between LAI and turbulent fluxes appear to contradict the experimental evidences. The dominance of the transpiration-driven over the observed albedo-driven effects might suggest that LSMs have the incorrect balance of these two processes. Such mismatches shed light on the limitations of our current understanding and process representation of the vegetation control on the surface energy balance and help to identify critical areas for model improvement.
BibTeX
@article{forzieri2018evaluating,
  author = {Giovanni Forzieri and Grégory Duveiller and Goran Georgievski and Wei Li and Eddy Robertson and Markus Kautz and Peter Lawrence and Lorea Garcia San Martin and Peter Anthoni and Philippe Ciais and Julia Pongratz and Stephen Sitch and Andy Wiltshire and Almut Arneth and Alessandro Cescatti},
  title = {{Evaluating the Interplay Between Biophysical Processes and Leaf Area Changes in Land Surface Models}},
  journal = {Journal of Advances in Modeling Earth Systems},
  year = {2018},
  doi = {10.1002/2018ms001284},
}
Figure