Details

Autor(en) / Beteiligte

• Viskari, Toni
• Shiklomanov, Alexey
• Dietze, Michael C
• Serbin, Shawn P
• Poulter, Benjamin

Titel

The influence of canopy radiation parameter uncertainty on model projections of terrestrial carbon and energy cycling

Ist Teil von

• PloS one, 2019-07, Vol.14 (7), p.e0216512-e0216512

Ort / Verlag

United States: Public Library of Science

Erscheinungsjahr

2019

Quelle

MEDLINE

Beschreibungen/Notizen

• Reducing uncertainties in Earth System Model predictions requires carefully evaluating core model processes. Here we examined how canopy radiative transfer model (RTM) parameter uncertainties, in combination with canopy structure, affect terrestrial carbon and energy projections in a demographic land-surface model, the Ecosystem Demography model (ED2). Our analyses focused on temperate deciduous forests and tested canopies of varying structural complexity. The results showed a strong sensitivity of tree productivity, albedo, and energy balance projections to RTM parameters. Impacts of radiative parameter uncertainty on stand-level canopy net primary productivity ranged from ~2 to > 20% and was most sensitive to canopy clumping and leaf reflectance/transmittance in the visible spectrum (~400-750 nm). ED2 canopy albedo varied by ~1 to ~10% and was most sensitive to near-infrared reflectance (~800-1200 nm). Bowen ratio, in turn, was most sensitive to wood optical properties parameterization; this was much larger than expected based on literature, suggesting model instabilities. In vertically and spatially complex canopies the model response to RTM parameterization may show an apparent reduced sensitivity when compared to simpler canopies, masking much larger changes occurring within the canopy. Our findings highlight both the importance of constraining canopy RTM parameters in models and valuating how the canopy structure responds to those parameter values. Finally, we advocate for more model evaluation, similar to this study, to highlight possible issues with model behavior or process representations, particularly models with demographic representations, and identify potential ways to inform and constrain model predictions.