Improving Representation of Deforestation Effects on Evapotranspiration in the E3SM Land Model

• Published in: Journal of advances in modeling earth systems 2019-08, Vol.11 (8), p.2412-2427
• Main Authors: Cai, Xitian, Riley, William J., Zhu, Qing, Tang, Jinyun, Zeng, Zhenzhong, Bisht, Gautam, Randerson, James T.
• Format: Article
• Language: English
• Subjects: Atmosphere; Atmospheric dynamics; Biogeochemistry; Carbon cycle; climate effects; Deforestation; Deforestation effects; Dynamics; E3SM; Earth; Earth system model; Energy; ENVIRONMENTAL SCIENCES; Evapotranspiration; Evapotranspiration models; Heat transfer; Hydrologic cycle; land use and land cover; Latent heat; Latent heat flux; Moisture content; Parameterization; Parameters; Photosynthesis; Runoff; Simulation; Soil; Soil dynamics; Soil water; Studies; Sustainability
• ISSN: 1942-2466; 1942-2466
• DOI: 10.1029/2018MS001551

Evapotranspiration (ET) plays an important role in land‐atmosphere coupling of energy, water, and carbon cycles. Following deforestation, ET is typically observed to decrease substantially as a consequence of decreases in leaf area and roots and increases in runoff. Changes in ET (latent heat flux) revise the surface energy and water budgets, which further affects large‐scale atmospheric dynamics and feeds back positively or negatively to long‐term forest sustainability. In this study, we used observations from a recent synthesis of 29 pairs of adjacent intact and deforested FLUXNET sites to improve model parameterization of stomatal characteristics, photosynthesis, and soil water dynamics in version 1 of the Energy Exascale Earth System Model (E3SM) Land Model (ELMv1). We found that default ELMv1 predicts an increase in ET after deforestation, likely leading to incorrect estimates of the effects of deforestation on land‐atmosphere coupling. The calibrated model accurately represented the FLUXNET observed deforestation effects on ET. Importantly, the search for global optimal parameters converged at values consistent with recent observational syntheses, confirming the reliability of the calibrated physical parameters. Applying this improved model parameterization to the globe scale reduced the bias of annual ET simulation by up to ~600 mm/year. Analysis on the roles of parameters suggested that future model development to improve ET simulation should focus on stomatal resistance and soil water‐related parameterizations. Finally, our predicted differences in seasonal ET changes from deforestation are large enough to substantially affect land‐atmosphere coupling and should be considered in such studies. Plain Language Summary Deforestation changes Earth's surface characteristics and affects the water cycle and climate. Although Earth system modeling is an important tool to understand the effects of deforestation, current models have large uncertainties. Here we used FLUXNET‐based observations to identify biases in representing deforestation effects on evapotranspiration (ET) in the Energy Exascale Earth System Model (E3SM). Results showed these biases are mostly associated with the representation of trees, not with smaller vegetation types (e.g., grasses). We then used the observations to optimize model parameters and improved simulations of ET and sensible heat fluxes following deforestation. Globally, these improvements led to a reduction in ET bias of 600