Assessment of a tiling energy budget approach in a land surface model, ORCHIDEE-MICT (r8205)

Yi Xi, Chunjing Qiu, Yuan Zhang, Dan Zhu, Shushi Peng, Gustaf Hugelius, Jinfeng Chang, Élodie Salmon, Philippe CiaisGeoscientific 2024 · DOI

TL;DR

This paper introduces a "tiling" approach to a land surface model (ORCHIDEE-MICT) that accounts for different vegetation types within a single grid cell. This improved representation leads to warmer surface and soil temperatures, wetter soils, and increased soil carbon storage across the Northern Hemisphere, with a significant enhancement in permafrost simulation.

What Problem Does This Paper Address?

Land surface models (LSMs) are crucial tools for understanding Earth's climate system, but they often struggle to represent the complex reality of landscapes. A key limitation is that a single grid cell in these models typically represents an average of different land cover types (like forests, grasslands, bare soil), ignoring the significant variations in how these types interact with energy, water, and carbon cycles. For instance, a forest and a grassland within the same area can have vastly different surface temperatures due to their unique properties like albedo (how much sunlight they reflect) and how they release water vapor. This "spatial heterogeneity" at scales smaller than the model's grid resolution can lead to inaccurate simulations of crucial processes, especially in regions with diverse vegetation and complex terrain, like the Arctic.

Methods

The authors implemented a "tiling energy budget" approach in the ORCHIDEE-MICT model. Instead of calculating energy budgets for the entire grid cell as a single unit, this new version separates these calculations for each "plant functional type" (PFT) present within that cell. This means that properties like surface roughness, albedo, and heat fluxes are now calculated independently for each PFT (e.g., bare soil, different tree types, grasses, peatland). These PFT-specific calculations then feed into the model's energy, hydrology, and carbon cycle modules, allowing for a more nuanced representation of sub-grid processes. The model was then run using different simulation protocols to compare the new "tiling" version (MICT-teb) against the original version (MICT).

Datasets

  • Climate Forcing Data: CRU-JRA v2.3 (used for transient simulations).
  • Land Cover Maps: Derived from the TRENDY dataset for 15 PFTs and a peat map from Xu et al. (2018).
  • Satellite-Derived Data for Evaluation:
    • MODIS Land Surface Temperature (LST) from MCD11C3 Version 6.1.
  • MODIS Albedo from MCD43C3 Version 6.1.
  • MODIS Latent Heat Flux (LE) from MOD16A2GF Version 6.1.
  • In Situ Observations for Evaluation:
    • Soil temperature data from 268 Russian meteorological station sites (Sherstiukov, 2012).
  • Permafrost datasets from Brown et al. (2002) and Obu et al. (2019).

Key Insights and Ideas

  • Heterogeneity Matters: The study strongly reinforces that accounting for sub-grid heterogeneity in surface properties and energy budgets is crucial for accurate climate modeling, particularly for processes like permafrost dynamics.
  • Permafrost Gains Big: The most significant improvement is in permafrost simulation. By better representing the varied thermal properties of different land cover types, the model can now simulate discontinuous permafrost areas, which was a major limitation before.
  • Energy Balances are Interconnected: Changes in surface energy budgets (driven by PFT-specific properties) cascade through to affect soil temperature, moisture, and ultimately carbon storage.
  • Model Limitations Persist: Despite improvements, the model still struggles with accurately simulating soil moisture and soil organic carbon, which impacts the overall temperature bias.

Key Results

  • Warmer and Wetter: The new tiling version (MICT-teb) generally shows warmer surface and soil temperatures (around +0.5°C, +3%), wetter soil moisture (around +10 kg m⁻², +2%), and increased soil organic carbon storage (around +170 Pg C, +9%) across the Northern Hemisphere compared to the original model.
  • Improved Permafrost Simulation: The model now accounts for discontinuous permafrost, covering approximately 3 × 10⁶ km², which is a significant advancement for permafrost-related research.
  • Reproduces Spatial Gradients: The model successfully reproduces observed spatial gradients in surface and soil temperatures.
  • Short-term vs. Long-term: Surface temperature and energy budgets respond quickly to the new tiling approach, reaching stability within a few years. Hydrology also responds relatively fast (up to 5 years), while soil carbon storage accumulates over longer timescales.
  • Model Performance: While the model captures spatial patterns of temperature well, it still exhibits biases, particularly in wet regions (overestimation) and dry areas (underestimation). The albedo and LE simulations also show some biases compared to satellite data.

Limitations and Future Work

  • Soil Moisture and SOC Uncertainties: The authors acknowledge that uncertainties in simulating soil moisture and soil organic carbon in the original ORCHIDEE-MICT model persist and prevent a complete elimination of temperature biases.
  • Prescribed vs. Simulated SOC: The use of simulated SOC for PFTs is an improvement, but it means that SOC formed by processes not simulated by the model (e.g., Pleistocene ecosystems) might be missed, leading to potential mismatches.
  • Data Limitations: Uncertainties in climate forcing data, prescribed peatland maps, and satellite data (e.g., differences in what LST represents) contribute to model-data discrepancies.
  • Future Work: The improved permafrost simulation capability is highlighted as a key advancement that will facilitate future permafrost-related studies. Further work could focus on improving the representation of soil moisture and SOC, and refining the parameterizations of processes like snow insulation. The authors also suggest that comparing tiling approaches across different land surface models can inspire future developments.

Summary updated

Abstract

The surface energy budget plays a critical role in terrestrial hydrological and biogeochemical cycles. Nevertheless, its highly spatial heterogeneity across different vegetation types is still missing in the ORCHIDEE-MICT (ORganizing Carbon and Hydrology in Dynamic EcosystEms–aMeliorated Interactions between Carbon and Temperature) land surface model. In this study, we describe the representation of a tiling energy budget in ORCHIDEE-MICT and assess its short-term and long-term impacts on energy, hydrology, and carbon processes. With the specific values of surface properties for each vegetation type, the new version presents warmer surface and soil temperatures (∼ 0.5 °C, +3 %), wetter soil moisture (∼ 10 kg m−2, +2 %), and increased soil organic carbon storage (∼ 170 Pg C, +9 %) across the Northern Hemisphere. Despite reproducing the absolute values and spatial gradients of surface and soil temperatures from satellite and in situ observations, the considerable uncertainties in simulated soil organic carbon and hydrological processes prevent an obvious improvement in the temperature bias existing in the original ORCHIDEE-MICT model. However, the separation of sub-grid energy budgets in the new version improves permafrost simulation greatly by accounting for the presence of discontinuous permafrost types (∼ 3×106 km2), which will facilitate various permafrost-related studies in the future.
BibTeX
@inproceedings{xi2024assessment,
  author = {Yi Xi and Chunjing Qiu and Yuan Zhang and Dan Zhu and Shushi Peng and Gustaf Hugelius and Jinfeng Chang and Élodie Salmon and Philippe Ciais},
  title = {{Assessment of a tiling energy budget approach in a land surface model, ORCHIDEE-MICT (r8205)}},
  booktitle = {Geoscientific},
  year = {2024},
  doi = {10.5194/gmd-17-4727-2024},
}
Figure