Representing Tall, Complex Canopies using Land Surface Models: Comparing CLM5 and CLM-ml to Sub-Canopy Data from a Costa Rican Rainforest

Tropical ecosystems are important contributors to the global cycling of water and carbon. Hence, accurate modeling of their land-surface processes is important to the understanding and prediction of land-atmosphere interactions. However, modeling evapotranspiration and photosynthesis, especially from mountainous vegetation, involves substantial errors. Our previous work has highlighted the issues related to photosynthesis parameters, but the lack of in-canopy variability, caused by overly simplistic model structure for sub-canopy layers and site-specific features (e.g., large/frequent rainfall, steep slope), can be also problematic. Here, we applied a multi-layered scheme to capture the site complexity and explore the implications at the stand/canopy scale.

This study examined the performance of the Community Land Model (CLM5) and Multi-Layered CLM (CLM-ml) against vertical profile measurements from a tropical premontane rainforest in Costa Rica. For further investigation, several canopy shapes were applied via a statistical distribution, and their "true" shape (more complex shape) was fitted using mixed-distributions. To reflect the canopy shapes in CLM-ml, a first-order-closure model was applied for turbulence.

The simulation results showed that the amplitude of diurnal variations for leaf temperature and leaf wetness were significantly improved compared to the single-layer model (CLM5), decreasing by 67% and 47%, respectively. With CLM-ml, different canopy shapes caused different net and photosynthetically active radiation profiles, which affects predictions of GPP and transpiration. Canopy shape also mattered for reproducing measured wind profiles. However, while this point-scale measurement of canopy shape gave a reasonable approximation for simulation and comparison to capture the feature of this complex terrain, the model with this point-scale canopy cannot yet sufficiently cover the spatial variability.

This study indicates that representing sub-canopy activity through structure improvement can produce more accurate estimations of energy and hydrological processes in CLM. However, more extensive monitoring of sub-canopies is necessary for further study and to increase model reliability for other sites with complex terrain and irregular vegetation roughness.